I build AI and backend systems that people can actually depend on.
Hi, I'm Akshit. I like working on the part that comes after the demo—when an idea has to survive messy data, unreliable inputs, changing requirements, and real users.
My background spans machine learning, backend engineering, and cloud infrastructure. I've worked on forecasting, recommendation systems, document intelligence, APIs, and deployment workflows. Lately, I've been especially interested in making AI systems easier to evaluate, observe, and trust.
I care about technology that is useful before it is impressive.
I'm currently a Software Developer at TOGA Technology, where I work across backend services, AWS infrastructure, and deployment automation.
Before that, I worked on forecasting and multimodal ML at Klippie, and on recommendation and NLP systems at Rejoicehub.
- Japanese OCR for RAG — a page-by-page pipeline for turning large Japanese PDFs into clean, traceable text without exhausting memory.
- AI Coach — an experimental retrieval-backed application for supportive mental-health conversations, built with FastAPI, Streamlit, and Pinecone.
- MCP experiments — early explorations in connecting local tools for messaging, voice, and task automation.
Some of my professional work lives behind private repositories. The projects above are smaller, public examples of how I think: understand the failure modes, keep the architecture explainable, and make the result usable.
Python, FastAPI, Node.js, PostgreSQL, AWS, Docker, and GitHub Actions—plus whatever simpler tool fits the problem better.
I hold an M.S. in Computer Science from New York Institute of Technology and a B.Tech. in Computer Engineering from UKA Tarsadia University.
If you're working on a difficult problem where AI has to meet a real system, I'd be glad to hear about it.