Who's Watching?
Recruiter
Developer
Explorer
Adventurer
Tejas Chakkarwar - Software Engineer
Master's Student @ San Jose State University
Dynamic and results-driven Software Engineer with 2+ years in full-stack development across high-impact, large-scale applications. Expertise in Java, Spring Boot, React, AWS, and microservices architecture.
Today's Top Picks for recruiter
Work Permit
F1 Visa • Valid until 2030
Skills
Spring Boot • React • AWS
Experience
Accelya • Hitachi Vantara
Projects
ResuMatch • RouteGuard • SaaS
Extracurriculars
ACM • Rotary International
Continue Watching for recruiter
Music
Interests & Hobbies
Reading
Books That Shaped My Journey
Blogs
Technical Writing
GitHub
View My Repositories
LeetCode
Coding Challenges
Contact Me
Get in Touch
Skills
Spring Boot • React • AWS
Programming Languages
Frameworks & Libraries
Data Stores & Messaging
Cloud & DevOps
Developer Tools
📅 Work Experience & Education Timeline
San Jose State University
Master's 🎓
🎓 Master of Science in Computer Science
- Coursework: Enterprise Distributed Systems, Cloud Computing, Machine Learning
- Focus on advanced software engineering and distributed systems
Software Engineer 💼
Accelya Solutions
🔧 Spring Boot, Microservices, REST APIs, SQL, Redis, Ehcache
- Migrated six legacy modules into Spring Boot microservices, reducing system latency by 29%
- Integrated over 12 REST and SOAP partner APIs, reducing reconciliation time by 25%
- Optimized 100+ SQL queries on 5M+ rows with indexing and batch processing
- Implemented asynchronous Java processing with CompletableFuture and caching
Big Data Analytics Intern 🎉
Hitachi Vantara
🔧 Python, PySpark, Pentaho, Hive, Snowflake, Power BI
- Migrated 50K+ lines of SAS code to Python/PySpark, reducing licensing costs by 30%
- Designed and maintained ETL pipelines for 100M+ record datasets
- Developed 5+ Power BI dashboards, reducing reporting turnaround time by 25%
- Collaborated with Agile teams to analyze data migration workflows
Savitribai Phule Pune University
Bachelor's 🎓
🎓 Bachelor of Computer Engineering with Data Science
- GPA: 3.71/4.00
- Coursework: Data Structures, Algorithms, Database Management, Machine Learning
- Led ACM Student Chapter as Secretary
- Volunteered with Rotary International teaching coding to underprivileged students
💻 Featured Projects
Building scalable solutions with modern technologies
ThinkFlow
⭐ GitHubAn advanced deep learning system that decodes raw EEG (electroencephalography) signals into coherent natural language text. The project implements and compares two state-of-the-art neural architectures for brain-computer interface applications, showcasing expertise in both modern Transformer-based models and classical sequence-to-sequence approaches.
- Transformer-Based Model (VQ-VAE + BART): Custom CNN encoder with 6 stacked Conv1D layers processing 105-channel EEG signals (5,500 timesteps), 8-head self-attention for long-range dependencies, Vector Quantization with 2,048-embedding codebook creating a "brain vocabulary" of 57 discrete tokens, and pretrained BART transformer for cross-modal translation
- LSTM Sequence-to-Sequence Model: Bidirectional 2-layer LSTM encoder with optional channel reduction (105→32), Bahdanau attention mechanism, 2-layer unidirectional LSTM decoder with attention-augmented inputs, ~15M trainable parameters optimized for EEG-to-text translation
- Dual Model Comparison: Side-by-side interface allowing users to compare Transformer vs LSTM performance on identical EEG inputs with real-time processing
- Interactive Visualization: Animated waveform displays, raw tensor data viewer with modal interface, color-coded output comparisons (generated vs expected)
- Advanced Features: Custom straight-through estimator for gradient flow through discrete codebook, configurable teacher forcing (0.5 ratio), Word Error Rate (WER) and word accuracy metrics, model performance metrics with confidence scores
- Modern UI/UX: Glassmorphism design with neural grid background effects, Framer Motion animations, responsive layout optimized for desktop and mobile
Sentinel
A production-ready, cloud-native email marketing platform built from scratch to solve scalability and reliability challenges. Fully serverless SaaS application that processes over 12,500 emails per minute and automatically scales based on demand.
- AI-Powered Automation: Generate complete email campaigns from prompts using Google Gemini with actionable performance insights
- Advanced Analytics Dashboard: Temporal patterns, device/browser/OS distribution, geographic insights, and link performance tracking
- Enterprise-Grade Reliability: 12.5x throughput improvement through Lambda concurrency optimization, exponential backoff with jitter
- Multi-Region Deployment: DynamoDB Global Tables replicate data across US, Europe, and Asia-Pacific regions in real-time
- Security First: HTML content sanitization prevents XSS attacks, URL validation blocks malicious links, API key authentication
- Event-Driven Architecture: Real-time event tracking for opens, clicks, bounces with asynchronous DynamoDB writes
CodeMedic
⭐ GitHubAn AI-powered automated bug fixing agent that analyzes Sentry errors, identifies problematic code, proposes fixes, tests them in a sandbox environment, and creates draft pull requests.
- Smart Error Analysis: Automatically analyzes Sentry error data including stack traces, error messages, and metadata
- AI-Powered File Discovery: Uses browser_use to intelligently navigate GitHub repositories and identify problematic files
- Intelligent Fix Generation: Leverages Google Gemini 2.5 Flash to propose comprehensive fixes with explanations
- Sandbox Testing: Tests fixes in isolated Daytona sandboxes before deployment
- Automated PR Creation: Creates draft pull requests with detailed descriptions
- Streamlit UI: User-friendly web interface for monitoring and controlling the bug fixing process
LiveAgent
⭐ GitHubA powerful multi-agent AI system that transforms your ideas into fully functional applications in minutes. LiveAgent uses CrewAI to orchestrate multiple specialized AI agents that collaboratively generate, test, and validate code for both frontend and backend components.
- Multi-Agent Code Generation: AI agents collaborate to generate production-ready code
- Integrated IDE: Built-in code editor with file management and terminal support
- Real-time Collaboration: Automerge-powered sync for collaborative editing
- Full-Stack Support: Generates both frontend (React/TypeScript) and backend code
- Progress Tracking: Real-time monitoring of code generation progress
- Specialized Agents: Orchestrator, Frontend, Backend, and QA agents work together
Stock Market Indices Tracker
A full-stack web application for tracking real-time stock market indices with historical data visualization. Built with Spring Boot backend and Next.js frontend, this application efficiently manages API rate limits through intelligent caching strategies.
- Real-time Index Tracking: Monitor live prices for SPY (S&P 500), DIA (Dow Jones), QQQ (NASDAQ-100), and IWM (Russell 2000)
- 30-Day Price History: Interactive charts showing historical price trends using Recharts
- Auto-refresh: Index prices update automatically every 90 seconds
- Rate Limit Management: Visual display of API usage with monthly and per-minute tracking
- Intelligent Caching: Redis caching with custom TTL strategies reduces API calls while maintaining data freshness
- Rate Limiting: Enforces 20 requests/minute and 500 requests/month limits
CuriosityAI
⭐ GitHubAI-powered agentic system that democratizes invention and R&D by identifying research gaps, analyzing feasibility, and generating research proposals with automated code pushing to GitHub.
- Identifies unexplored innovations using GMM/KDE density estimation on embeddings from ArXiv/PubChem APIs
- Autonomous agents evaluate novelty, technical viability, and ethical considerations using LLMs
- Generates structured research proposals with abstracts, methods, and impact assessments via RAG
- Auto-commits prototype code to GitHub repositories via GitHub API integration
ResuMatch
⭐ GitHubAI-powered resume optimization platform that surfaces top-match roles and generates ATS-optimized resumes using agentic workflows.
- Orchestrated agentic loop with LangGraph/LangChain + Google Gemini for iterative resume rewriting
- Built REST APIs with JSON guardrails and CORS-enabled integration with external ML services
- Implemented lightweight RAG layer using in-memory knowledge base for company-specific patterns
- Created job-matching system with relevance scoring and visual progress tracking
RouteGuard
⭐ GitHubFull-stack transport order management system with role-based interfaces for customers, drivers, and business clients.
- Implemented JWT-based authentication with Spring Security for 100+ mock users
- Automated invoice generation using Spring Boot schedulers
- Built role-based access control for secure workflow management
- Deployed on AWS for cloud integration and scalability
SaaS Subscription Management
⭐ GitHubEnterprise-grade backend service for managing SaaS subscriptions, billing, and user management with automated workflows.
- Built comprehensive user management with JWT authentication and role-based access control
- Integrated Stripe for payment processing with automated billing cycles
- Implemented Redis caching for session management and rate limiting
- Automated invoice generation with PDF creation and S3 storage
- Added RabbitMQ for asynchronous processing of heavy operations
Hotel Management System
⭐ GitHubDesktop application for hotel staff to manage guest records, bookings, and billing with secure authentication.
- Built MySQL backend with role-based authentication for secure data handling
- Designed intuitive Tkinter GUI for efficient guest management
- Implemented automated reporting features for business analytics
- Optimized SQL queries to improve system efficiency
🌟 Beyond the Code
Leadership, community engagement, and personal growth through diverse experiences
Leadership & Student Organizations
Secretary
ACM Student Chapter
- Orchestrated operations for 200+ member chapter with 15+ technical events
- Coordinated workshops and competitions hosting 50-100 attendees each
- Bridged communication between executive board, faculty, and members
Alumni Relations Team Member
Indian Student Organization
- Foster connections across 100+ member alumni network
- Coordinate networking events and cultural programs
- Support new students' transition to campus life with mentorship
Active Member
Google Developers Student Club
- Supported 10+ hackathons, coding competitions, and tech talks
- Promoted club activities and recruited new members
- Contributed to community-building initiatives
Community Service & Volunteering
Education Volunteer
Rotary International
- Provided educational support to 25+ underprivileged students during COVID-19
- Taught Math and Coding through virtual platforms
- Created culturally appropriate teaching materials for diverse learners
- Managed weekly schedules while adapting to changing circumstances
Event Coordination
Event Coordinator
Debate & Elocution Competitions
- Organized intercollegiate competitions for 100+ participants
- Coordinated with 5+ colleges for promotion and outreach
- Managed registrations, scheduling, and participant communications
- Handled event logistics including setup and equipment management
Sports & Athletics
District Level Table Tennis Player
Competitive Sports
- Competed in district-level table tennis tournaments
- Maintained rigorous training schedule alongside full-time academics
- Developed mental resilience and adaptability through competitive sports
- Built teamwork skills through collaboration with coaches and teammates
📚 Books That Shaped My Journey
These books have influenced my perspectives, motivation, and self-growth.
Atomic Habits
A practical guide to building good habits and breaking bad ones.
Rich Dad Poor Dad
An eye-opener on wealth, assets, and financial literacy.
The Alchemist
A magical journey of following one's dreams.
Eat That Frog
A motivational book on overcoming procrastination.
Deep Work
Rules for focused success in a distracted world.
The Pragmatic Programmer
Essential reading for software developers.
Clean Code
A handbook of agile software craftsmanship.
Verity
A gripping psychological thriller that keeps you on edge.
🎵 Music
Discover my musical journey and favorite tracks
Explore by Genre
🎧 EDM (Electronic Dance Music)
Gold Skies (EP)
Martin Garrix (2014)
Includes hits like Animals and Wizard that shaped modern progressive house.
True
Avicii (2013)
A genre-blending album mixing EDM with folk and soul (Wake Me Up, Hey Brother).
Motion
Calvin Harris (2014)
Mainstream EDM bangers (Summer, Blame, Outside).
4×4=12
Deadmau5 (2010)
Progressive and electro-house classic featuring Ghosts 'n' Stuff.
Until Now
Swedish House Mafia (2012)
Anthemic festival energy (Don't You Worry Child).
🎸 Folk Rock / Indie Folk
Sigh No More
Mumford & Sons (2009)
Emotional harmonies and banjo-driven folk anthems (Little Lion Man).
Fleet Foxes
Fleet Foxes (2008)
Rich vocal layering and poetic lyrics defining indie folk.
For Emma, Forever Ago
Bon Iver (2007)
A minimal, heartfelt masterpiece written in isolation.
The Lumineers
The Lumineers (2012)
Accessible folk storytelling (Ho Hey, Stubborn Love).
My Head Is an Animal
Of Monsters and Men (2011)
Icelandic indie folk glory (Little Talks).
🎤 Hip-Hop / Rap
To Pimp a Butterfly
Kendrick Lamar (2015)
A cultural landmark blending jazz, funk, and rap — Pulitzer-level storytelling.
My Beautiful Dark Twisted Fantasy
Kanye West (2010)
Lavish, genre-defying production and introspection.
2014 Forest Hills Drive
J. Cole (2014)
Authentic, soulful, and self-reflective — no features, pure vision.
ASTROWORLD
Travis Scott (2018)
Trap-driven, psychedelic production (SICKO MODE).
Take Care
Drake (2011)
Emotional hip-hop infused with R&B (Marvins Room, Headlines).
🕊️ Sufi
Shahenshah
Nusrat Fateh Ali Khan (1989)
His vocal range and spiritual power are unmatched.
Raqs-e-Bismil
Abida Parveen (2001)
Deeply spiritual qawwalis evoking transcendence.
Kailasa
Kailash Kher (2006)
A modern Sufi-pop fusion with songs like Teri Deewani.
Azadi
Junoon (1997)
Pioneers of Sufi rock blending Pakistani poetry with guitars (Sayonee).
Coke Studio Pakistan
Various Artists - Season 3 & 4
Iconic Sufi renditions like Alif Allah (Jugni) and Tajdar-e-Haram.
☁️ Lo-Fi / Chillhop
beats to relax/study to
ChilledCow (Lofi Girl)
The most iconic lo-fi stream that defined the genre.
Hiraeth
Idealism (2016)
Smooth, dreamy instrumentals that embody nostalgic calm.
Life
Jinsang (2016)
A classic lo-fi hip-hop album with warm, jazzy undertones.
Modal Soul
Nujabes (2005)
Lo-fi's spiritual father; blends hip-hop with Japanese jazz (Luv(sic) series).
Beat Tapes, Vol. 1–3
eevee (2015–2017)
Gentle, minimal beats for focus and comfort.
Tejas Chakkarwar
Software Engineer
With 2+ years in full-stack development, skilled in Java, Spring Boot, React, AWS, and microservices architecture. Passionate about building scalable systems and solving complex problems.
San Jose State University | Accelya Solutions
I'm always up for a chat or a coffee! Feel free to reach out.
👨💻 Developer Hub
Live GitHub stats, quick links, and contributions at a glance.
GitHub Activity
- Total Repositories: N/A
- Total Commits (2025): N/A
- Languages: Java, Python, JavaScript, Go
- Active Projects: 3+
Contributions
🚀 Featured Project - Sentinel
Serverless, event-driven email marketing platform with real-time analytics and multi-region resilience.
Sentinel
A production-ready, cloud-native email marketing platform built to solve scalability and reliability challenges. Fully serverless architecture processes high-volume campaigns with real-time analytics and robust failure handling.
- Event-driven pipelines with SQS + Lambda for reliable processing and retries
- Multi-region DynamoDB Global Tables for low-latency global reads/writes
- Throughput optimized to 12.5k emails/min with concurrency tuning and backoff
- Security: HTML sanitization, URL validation, API key auth, least-privilege IAM
- Observability: structured logging, metrics, and alerting for delivery health
- Analytics: device/browser/geo insights, link performance tracking
🏗️ Architecture Highlights
1) Lambda Function Structure
Environment wiring for campaigns, segments, events, and queue targets.
# infra/modules/lambdas/main.tf
resource "aws_lambda_function" "start_campaign" {
function_name = "${var.name}-start-campaign"
role = var.roles.lambda_exec
handler = local.lambda_handler
runtime = local.lambda_runtime
filename = "${path.module}/.artifacts/start_campaign.zip"
source_code_hash = filebase64sha256("${path.module}/.artifacts/start_campaign.zip")
timeout = local.timeout_long
environment {
variables = {
DYNAMODB_CAMPAIGNS_TABLE = var.dynamodb_campaigns_table
DYNAMODB_SEGMENTS_TABLE = var.dynamodb_segments_table
DYNAMODB_EVENTS_TABLE = var.dynamodb_events_table
SEND_QUEUE_URL = var.queues.send_queue_url
AB_TEST_ANALYZER_LAMBDA_ARN = aws_lambda_function.ab_test_analyzer.arn
EVENTBRIDGE_ROLE_ARN = var.scheduler_invoke_role_arn
}
}
}# services/send_worker/handler.py
def lambda_handler(event, _context):
# Triggered by SQS event. Each record has:
# {"campaign_id":123, "recipient_id":456, "email":"user@example.com"}
for rec in event.get("Records", []):
body = json.loads(rec["body"])
campaign_id = body["campaign_id"]
recipient_id = body["recipient_id"]
email = body["email"]
variation_id = body.get("variation_id")
# Process email sending logic...2) SQS Message Processing
Concurrency-capped worker and SQS fan-out respecting SES limits (≈14 emails/sec).
# infra/modules/lambdas/main.tf
resource "aws_lambda_event_source_mapping" "send_worker_sqs" {
event_source_arn = var.queues.send_queue_arn
function_name = aws_lambda_function.send_worker.arn
batch_size = local.ses_batch_size
maximum_batching_window_in_seconds = local.ses_batching_window_seconds
# Throughput = ses_max_concurrency × ses_batch_size = 2 × 7 = 14 emails/sec
scaling_config {
maximum_concurrency = local.ses_max_concurrency
}
}# services/start_campaign/handler.py
# Fan out to SQS in batches of up to 10 messages
for batch in _chunks(contacts, 10):
entries = []
for c in batch:
message_body = {
"campaign_id": campaign_id,
"recipient_id": c["id"],
"email": c["email"],
"template_data": {
"subject": campaign.get("email_subject", campaign.get("subject", "")),
"html_body": campaign.get("email_body", campaign.get("html_body", "")),
"from_email": campaign.get("from_email", "noreply@thesentinel.site"),
"from_name": campaign.get("from_name", "Sentinel")
}
}
entries.append({"Id": str(c["id"]), "MessageBody": json.dumps(message_body)})
sqs.send_message_batch(QueueUrl=SQS_URL, Entries=entries)3) DynamoDB Schema Design
GSIs for owner, campaign, and recipient lookups; TTL for link cleanup.
# infra/modules/dynamodb/main.tf (selected)
resource "aws_dynamodb_table" "users" {
name = "${var.name}-users"
billing_mode = "PAY_PER_REQUEST"
hash_key = "id"
attribute { name = "id" type = "S" }
attribute { name = "email" type = "S" }
attribute { name = "api_key" type = "S" }
global_secondary_index { name = "email_index" hash_key = "email" projection_type = "ALL" }
global_secondary_index { name = "api_key_index" hash_key = "api_key" projection_type = "ALL" }
}
resource "aws_dynamodb_table" "campaigns" {
name = "${var.name}-campaigns"
billing_mode = "PAY_PER_REQUEST"
hash_key = "id"
attribute { name = "id" type = "S" }
attribute { name = "owner_id" type = "S" }
global_secondary_index { name = "owner_index" hash_key = "owner_id" projection_type = "ALL" }
}
resource "aws_dynamodb_table" "events" {
name = "${var.name}-events"
billing_mode = "PAY_PER_REQUEST"
hash_key = "id"
attribute { name = "id" type = "S" }
attribute { name = "campaign_id" type = "S" }
global_secondary_index { name = "campaign_index" hash_key = "campaign_id" projection_type = "ALL" }
}
resource "aws_dynamodb_table" "link_mappings" {
name = "${var.name}-link-mappings"
billing_mode = "PAY_PER_REQUEST"
hash_key = "tracking_id"
attribute { name = "tracking_id" type = "S" }
attribute { name = "campaign_id" type = "S" }
attribute { name = "recipient_id" type = "S" }
global_secondary_index {
name = "campaign_recipient_index"
hash_key = "campaign_id"
range_key = "recipient_id"
projection_type = "ALL"
}
ttl { attribute_name = "expires_at" enabled = true }
}4) SQS + Retry & DLQ
Queue with DLQ and exponential backoff retry helper.
# infra/modules/queues/main.tf
resource "aws_sqs_queue" "dlq" {
name = "${var.name}-dlq"
message_retention_seconds = 1209600 # 14 days
}
resource "aws_sqs_queue" "send_queue" {
name = "${var.name}-send-queue"
visibility_timeout_seconds = 60
receive_wait_time_seconds = 20 # long polling
redrive_policy = jsonencode({
deadLetterTargetArn = aws_sqs_queue.dlq.arn,
maxReceiveCount = 3 # fail faster
})
}# services/common.py
def exponential_backoff_retry(func, max_retries=3, base_delay=1.0, max_delay=60.0,
exponential_base=2, jitter=True, retryable_exceptions=(Exception,)):
"""Execute a function with exponential backoff retry logic"""
last_exception = None
for attempt in range(max_retries + 1):
try:
return func()
except retryable_exceptions as e:
last_exception = e
if not is_retryable_error(e) or attempt >= max_retries:
raise
delay = min(base_delay * (exponential_base ** attempt), max_delay)
if jitter:
delay = delay * (0.5 + random.random())
time.sleep(delay)⚡ Performance Optimizations
- Exponential backoff with jitter to respect SES rate limits
- Batch processing for DynamoDB writes (25 items per batch)
- CloudFront caching for static dashboard assets
- Connection pooling for database operations
🔧 Infrastructure as Code
- Terraform modules for reproducible deployments
- CI/CD pipeline with GitHub Actions
- Blue-green deployment strategy
- Automated rollback on CloudWatch alarms
📊 Monitoring & Observability
- CloudWatch metrics for Lambda invocations, errors, duration
- X-Ray tracing for distributed request tracking
- Custom metrics for email delivery rates
- SNS alerts for critical failures
🔄 SJSU RideShare - Ongoing Project
Production-grade, scalable carpooling platform for San José State University students, faculty, and staff.
📊 Project Overview
SJSU RideShare is a production-grade, scalable carpooling platform designed exclusively for San José State University students, faculty, and staff. The platform enables secure ride-sharing within the campus community, reducing transportation costs and carbon footprint while fostering connections among students.
🏗️ System Architecture
Architecture Layers
👤 User Service (8001)
- Authentication & Authorization (JWT)
- User Profile Management
- Driver License Verification
- Rating System
- SJSU Email Validation (@sjsu.edu)
🚗 Ride Service (8002)
- Ride CRUD Operations
- Advanced Search (proximity, time, price)
- Google Maps Integration
- Route Optimization
- Recurring Rides
📅 Booking Service (8003)
- Booking State Machine (Pending → Approved → Completed)
- Seat Reservation with Pessimistic Locking
- Payment Processing (Stripe)
- Cancellation Policy Engine
- Refund Management
🔔 Notification Service (8004)
- Multi-channel Notifications (Email, Push, SMS)
- Email Templates (SendGrid)
- Firebase Cloud Messaging
- Notification Preferences
- Event-driven Architecture
📍 Tracking Service (8005)
- Real-time WebSocket Connections
- Location Broadcasting (Redis PubSub)
- Geofencing (500m approaching, 100m arrived)
- Dynamic ETA Calculation
- Route History Storage
🎯 Smart Matching Algorithm
Multi-Factor Scoring System (0-100 points) that intelligently matches passengers with drivers based on route compatibility, time match, price competitiveness, driver reputation, and user preferences.
⚙️ Backend Stack
📱 Frontend Stack
☁️ Infrastructure
🔌 External APIs
📈 Scalability & Infrastructure Strategy
Phase 1: MVP (Month 0-2)
- Railway / Render (PaaS)
- Managed PostgreSQL (1GB RAM)
- Upstash Redis (free tier)
- Sentry free tier
Phase 2: Growth (Month 3-4)
- Migration to AWS
- ECS Fargate containers
- RDS PostgreSQL (db.t3.medium)
- ElastiCache Redis
- Auto-scaling (2-10 containers)
Phase 3: Scale (Month 5-6)
- Kubernetes (EKS)
- PostgreSQL Master-Replica
- Redis Cluster (6 nodes)
- CloudFront CDN
- 99.9% uptime target
Phase 4: Multi-Region (Month 7-8)
- Active-Active multi-region
- Route 53 Geolocation
- Cross-region replication
- Kafka event streaming
- 99.95% uptime target
✨ Key Features & Technical Highlights
🔒 Security
- JWT Tokens (RS256 algorithm)
- Token Rotation & Refresh
- SJSU Email Verification
- Encryption at Rest (AES-256)
- TLS 1.3 in Transit
- PCI Compliance via Stripe
⚡ Performance
- Multi-level Caching (L1/L2/L3)
- Database Partitioning
- Connection Pooling
- Asynchronous Processing (Celery)
- API Rate Limiting (100 req/min)
- Response Compression (Gzip)
📊 Observability
- Prometheus + Grafana
- ELK Stack Logging
- Sentry Error Tracking
- CloudWatch Metrics
- X-Ray Tracing
- Custom Business Metrics
🧪 Testing
- 80%+ Test Coverage Target
- Unit Tests (pytest)
- Integration Tests
- E2E Tests (Detox)
- Load Testing (Locust/k6)
- Security Testing (OWASP)
💳 Payment Flow (Escrow Pattern)
⚡ Real-Time Architecture
WebSocket-based real-time location tracking with Redis PubSub for horizontal scaling. Updates every 5 seconds with location, speed, bearing, ETA, and geofence events.
- WebSocket connections with sticky sessions (Kubernetes)
- Redis PubSub for broadcasting to all passengers
- Connection pool management
- Reconnection with exponential backoff
- Geofencing: 500m approaching, 100m arrived notifications
- Dynamic ETA calculation based on real-time traffic
🗄️ PostgreSQL Schema
- Users: Indexed by email, phone, stripe_customer_id
- Rides: GiST index on origin_coords, partitioned by departure_time
- Bookings: Indexed by ride_id, passenger_id, status
- Partitioning: Time-based (monthly/quarterly)
- Row-level Locking: SELECT FOR UPDATE for seat reservation
- Connection Pool: 50 connections, PgBouncer
⚡ Redis Architecture
- Cluster: 6 nodes (3 master, 3 replica)
- Sentinel: Automatic failover
- Session Tokens: TTL 24h
- Geocoding Cache: TTL 30 days
- Route Cache: TTL 1 hour
- Location Data: TTL 24h, PubSub for real-time
🎓 Key Learnings & Technical Challenges
Microservices Complexity
Challenge: Managing inter-service communication and data consistency
Solution: Event-driven architecture with Redis PubSub, transaction boundaries, saga pattern
Real-time Scalability
Challenge: WebSocket connections don't scale horizontally
Solution: Redis PubSub broadcasting, sticky sessions, connection state management
Payment Security
Challenge: PCI compliance, refunds and disputes
Solution: Stripe for payment handling, escrow pattern, automated refund calculation
Geospatial Performance
Challenge: Fast proximity search with millions of coordinates
Solution: PostGIS with GiST indexing, Redis geospatial commands, aggressive caching
Race Conditions
Challenge: Multiple passengers booking last seat simultaneously
Solution: Pessimistic locking (SELECT FOR UPDATE), atomic operations, idempotent APIs