I design and deploy robust, scalable AI systems in real-world production environments. My work spans the full ML lifecycle — data engineering, model development, inference optimization, and production monitoring. I operate at the intersection of research and engineering, where accuracy, efficiency, and reliability are non-negotiable.

Machine Learning Engineer — Sujanix Pvt. Ltd., Bengaluru (2026 – Present)

Working on production-level computer vision systems. Improved analog meter dial detection accuracy by 7 percentage points in production, bringing the deployed system to 94% accuracy. Responsible for scalable inference pipelines, data workflow architecture, and end-to-end model reliability in live environments.

Production-level academic companion built for AP and TS engineering students. Target: 1000 students. Infrastructure cost: zero.

Students authenticate via Google, upload marksheet images, and the system extracts subject scores automatically through a PaddleOCR + Groq LLM pipeline engineered to handle messy OCR output, handwritten elements, and the varied marksheet formats across AP and TS colleges. The platform tracks performance over time, surfaces an AI chat companion that reasons directly over each student's actual marks, delivers personalized push notifications for exams and motivation, and scrapes board announcements in real time. Every constraint is verified against free tier limits. Designed to run at full capacity without spending a rupee.

Python FastAPI PaddleOCR Groq LLaMA 3.1 PostgreSQL Redis Cloudinary React Native Expo Google OAuth2 APScheduler SQLAlchemy Alembic JWT

Self-hosted intelligence tool for engineers who cannot afford to follow 6+ sources daily just to stay current.

You define your stack. NoiseCut ingests Hacker News, GitHub Trending, ArXiv, and Dev.to on a configurable schedule, passes every item through an LLM relevance pipeline scored against your profile, and surfaces only what matters — tool updates, breaking changes, research papers, and trending repos in your domain. Semantic search over all ingested history. Breaking change alerts pushed via WebSocket in real time. Daily digest delivered via Telegram bot. Swap between local Ollama and Groq with a single environment flag. Zero cost. Fully open source. Runs on CPU.

Python FastAPI LangChain Ollama Mistral 7B Q4 Groq ChromaDB sentence-transformers PostgreSQL Redis APScheduler Docker Nginx React Vite WebSockets

Machine Learning Engineer — Sujanix Pvt. Ltd. (2026 – Present) Designing, training, and deploying ML models for production systems. Building scalable inference pipelines and optimized data workflows. Owning model performance, monitoring, and continuous optimization in live environments.

Founder — SpaceDrift (Aug 2024 – Dec 2025) Founded and operated SpaceDrift, an MSME-registered sole proprietorship. Delivered end-to-end engineering project builds, professional data annotation services, and research support to PhD scholars. Worked directly with graduate researchers on ML-adjacent problem framing, dataset curation, and experimental pipelines.

Data Science Intern — BrainOvision Solutions Pvt. Ltd. (Feb 2024 – Apr 2024) Improved forecasting accuracy by 15% through structured feature engineering. Conducted advanced EDA and statistical modeling for real business use cases.

Machine Learning Intern — Indian Servers (Apr 2023 – Jun 2023) Built BERT-based NLP pipelines for sentiment and emotion classification. Improved model accuracy by 20% over baseline.

Languages

ML / DL

LLMs & GenAI

Vector DB & Embeddings

MLOps & Infra

Backend & Data

Tools

Machine Learning Specialization — DeepLearning.AI / Stanford University Data Science Using Python — NPTEL / IIT IBM SQL Fundamentals — Coursera

Open to collaborating on applied machine learning, deep learning systems, and production ML engineering. If the problem requires strong engineering discipline and precise ML, reach out.