About
I'm a passionate Software Engineer and Data Scientist currently working as a Forward Deployed Engineer at First Distributors, after finishing my Master's in Data Science at Stony Brook University. With a strong foundation in Computer Science from IIT Guwahati, I specialize in building scalable AI-powered solutions and modern web applications.
My experience spans from architecting security and compliance platforms at PTC and enterprise-grade AI document assistants at Oracle to building semantic search engines and working on cutting-edge ML research. I'm particularly passionate about the intersection of AI, web development, and user experience.
When I'm not coding, you can find me exploring the outdoors through , hiking, and . I believe in continuous learning and enjoy tackling complex problems that push the boundaries of technology.
Experience
Apr 2026 — Present As a Forward Deployed Engineer reporting directly to the CEO, I operate as a full-stack, one-person product team, owning frontend, backend, AI integration, and DevOps end to end. My flagship project is an internal RFQ-to-quotation automation platform that ingests incoming customer emails, extracts line items, and matches them against a ~100K-product catalog using a hybrid deterministic + LLM pipeline to auto-generate quotes. I secured the React/FastAPI app with Azure AD (MSAL, PKCE), integrated Microsoft Graph for multi-mailbox email ingestion and QuickBooks Online for financial data, and ran AI tooling in a privacy-first, no-egress configuration to meet customer data requirements.
- Python
- FastAPI
- React
- PostgreSQL
- OpenAI
- MCP
- Microsoft Graph API
- Azure AD
- Docker
Aug 2024 — Apr 2026 I architected scalable backend infrastructure on Azure and GCP for security and compliance software, supporting Windows, macOS, and Ubuntu platforms. I built high-availability web applications using Next.js, TypeScript, and Postgres, developed cross-platform desktop components with Go and Electron, and set up robust CI/CD, monitoring, and logging systems to ensure secure, reliable releases in a fast-paced startup environment while working closely with leadership to deliver customer-focused, security-first features.
- Azure
- Next.js
- TypeScript
- Postgres
- Go
- Prometheus
- Jenkins
Aug 2024 — May 2025 Engineered an AI-driven platform to generate and update 3D scene graphs using outputs from Grounded SAM and LLMs. Integrated FastAPI for LLM inference and designed CI/CD pipelines, improving semantic accuracy by 25% and reducing model response time by 40%.
- Python
- FastAPI
- LLMs
- Docker
- Kubernetes
- Jenkins
Dec 2024 — Jan 2025 Led development of Thalmaar AI's official website with scalable, responsive design. Built dynamic SEO-optimized pages and deployed a custom LLM-based chatbot, implementing comprehensive CI/CD pipelines and testing frameworks.
- React
- Next.js
- TailwindCSS
- FastAPI
- MongoDB
- GitHub Actions
- Jest
July 2022 — Aug 2023 Built enterprise-grade AI-powered document assistant achieving 95%+ query resolution rate. Engineered fraud detection system with Spring Boot and Kafka, designed secure RBAC-enabled REST APIs, and containerized microservices improving release frequency by 50%.
- GenAI
- FastAPI
- React
- Oracle Vector DB
- Spring Boot
- Kafka
- Docker
- Jenkins
May 2020 — Jul 2022 Worked in a small applied AI team to improve SaaS platforms for customer intelligence. I built scalable apps and APIs using Python, FastAPI, and JavaScript, managed deployments with Airflow and Docker, and deployed services on AWS with proper monitoring and logging. I also worked with Postgres, MongoDB, SQL, and Power BI, gaining full end-to-end product development experience.
- Python
- FastAPI
- JavaScript
- AWS
- Docker
- Postgres
- MongoDB
- SQL
- Power BI
Projects
Jan 2025 — Mar 2025 Built a semantic search engine with scalable web interfaces using React and Next.js. Designed high-performance FastAPI backend with <1s latency, integrated custom NLP embeddings with LLaMA LLM enhancing retrieval precision by 40%, and built fault-tolerant data ingestion pipeline processing 2.6M+ research papers.
- RAG
- LLM
- Python
- FastAPI
- React
- Next.js
- MongoDB
- Docker
- Kubernetes
- Jenkins
- BigQuery
Sep 2024 — Dec 2024 Developed a robust music recommendation system using the Spotify dataset by applying advanced preprocessing techniques such as binning, one-hot encoding, and PCA for dimensionality reduction, which improved model performance. Designed and evaluated four approaches—Random, Baseline, Content-Based, and Collaborative Filtering—using metrics like RMSE, Precision, and Catalog Coverage, with Collaborative Filtering achieving the highest accuracy. Successfully addressed challenges such as the cold start problem to enhance user experience.
- Python
- Pandas
- Scikit-learn
- Matplotlib/Seaborn
- Jupyter
- Spotify API
- Numpy
- KNN
- PCA
- Binning
- One-hot Encoding
Playground
Proof that I said "when I'm not coding" and then coded some games anyway. Five little arcade games — no backend, no accounts, just your browser.
Prefer full screen? Open the arcade
Psst — "fishing" and "running" in the About section are clickable.