I build high-performance, containerized backend services, microservices APIs, and stateful multi-agent AI systems. Combining Data Science with a Minor in Finance & Economics from IIT Madras to bridge technical architecture and business domain logic.
Proficient in Python and FastAPI, building low-latency RESTful APIs, data validation using Pydantic, and automated migrations using SQLAlchemy and Alembic.
Hands-on production experience designing stateful multi-agent pipelines (LangGraph/LangChain) and Retrieval-Augmented Generation services with Qdrant and Pinecone.
Experienced in containerizing workloads using Docker and orchestration queues (Celery/Redis, SQS/DynamoDB) to run non-blocking background workers.
Leveraging a minor in Finance & Economics from IIT Madras to build business-critical backend applications like transactional ledgers, e-commerce, and ERP systems.
Specializing in systems software and financial data analysis. Completed core coursework in Database Management Systems (DBMS), Cloud Computing, Software Testing, Deep Learning, and an academic Minor in Finance & Economics.
Optimized tutoring system delivering dynamic content generation (flashcards, quiz Q&As, 3D models) with Server-Sent Events (SSE) streaming and AWS queue ingestion.
An autonomous agentic workflow crawling job boards (Lever, Ashby) via headless Playwright, performing semantic match-ranking via PostgreSQL pgvector, and generating resumes asynchronously.
A conversational banking platform featuring OCR-based KYC parsing, real-time machine learning transaction fraud-detection (Scikit-Learn), and secure FastAPI endpoints.
Finetuned Google’s Gemma 3 1B LLM for sentiment classification across 13 Indian languages. Implemented LoRA (PEFT) and 4-bit quantization to optimize training parameters.
Developed a LightGBM regression pipeline on 50+ categorical/numerical features of clickstream behavior, tuning hyperparameters via Optuna's Bayesian search.
A multi-user occupancy tracking platform featuring an administration dashboard built with Vue.js, showing real-time slot statuses and check-in history charts.
A multi-user exam preparation portal built with Flask and HTML/CSS/JS, enabling administrators to manage course questions and students to take mock tests.
Designed a binary classification ensemble model (XGBoost/LightGBM) to predict malware infection probability based on high-dimensional computer telemetry.
Conducted data cleaning, product affinity market basket analysis, ABC Pareto categorization, and volume vs. margin strategic analysis on 9,500+ physical transactions.
I am actively seeking full-time opportunities or technical collaborations in Backend Development and Applied Generative AI. Whether you want to discuss a project, query my work experience, or just say hello, my inbox is open!