• Built a scalable recommendation backend using Python, FastAPI, and Django, serving 2M+ users with <200ms p95 latency at 10k+ RPS.

• Deployed services on Kubernetes-based distributed infrastructure, enabling auto-scaling, high availability, and fault tolerance.

• Developed hybrid recommendation systems (collaborative + content-based) with LLM embeddings and vector search (FAISS, Pinecone), improving cold-start accuracy by 35%

• Designed and delivered RESTful APIs and async GraphQL services with 99.99% uptime under high traffic

• Optimized PostgreSQL with sharding, indexing, and Redis caching to handle 5M+ daily events efficiently

• Built CI/CD pipelines (GitHub Actions, Jenkins) for automated testing, containerization, and Kubernetes deployments

• Created ETL pipelines using PySpark and SQL to process large-scale user interaction data

• Implemented real-time streaming with Kafka and Spark Streaming for instant recommendation updates

• Developed async systems with Celery, RabbitMQ, and Kafka for background processing and model retraining

• Used MLflow for A/B testing and model tracking, improving CTR, retention, and engagement

• Achieved 95%+ test coverage using PyTest for reliable deployments

• Designed highly scalable systems with rate limiting, retries, circuit breakers, handling up to 50k RPS

• Set up monitoring and observability (Prometheus, Grafana, ELK) and built dashboards using React & TypeScript

• Ensured secure and compliant systems with OAuth2, JWT, GDPR, and SOC2 standards

• Engineered and scaled internal AI developer tools using Python, streamlining model development, testing, and deployment

workflows, improving developer productivity by 35%.

• Built and productionized RESTful APIs and GraphQL services using FastAPI and Flask, enabling flexible, real-time AI/ML inference and seamless integration across internal financial platforms.

• Architected high-performance async and distributed systems using FastAPI, Celery, and RabbitMQ, supporting low-latencyinference and scalable background processing for compute-intensive workloads.

• Designed scalable microservices architecture and leveraged Flask for rapid prototyping, accelerating feature development cycles and system modularity.

• Developed end-to-end ML pipelines and reusable SDKs covering data preprocessing, feature engineering, model training, evaluation, benchmarking, and experiment tracking (MLflow), improving reproducibility and experimentation speed.

• Implemented LLM-powered features (prompt engineering, embeddings, semantic search) and built RAG pipelines, enabling context-aware, intelligent responses over large-scale document corpora.

• Deployed scalable AI/ML systems on AWS (S3, EC2, Lambda, API Gateway), optimizing for cost, performance, and highavailability in cloud-native environments.

• Designed and implemented secure, scalable backend architectures with OAuth2/JWT authentication, rate limiting, caching, and horizontal scaling to support high-throughput financial systems.

• Containerized applications using Docker and orchestrated deployments with Kubernetes, enabling auto-scaling, fault tolerance, and high availability.

• Built and optimized data storage and caching layers using PostgreSQL, MongoDB, and Redis, ensuring low-latency access and high-throughput data processing.

• Processed large-scale financial datasets using Apache Spark and PySpark and integrated real-time streaming with Kafka, enabling continuous ingestion and near real-time analytics.