Akritah Sahu

I'm a third-year Computer Science student (AI & ML specialization) at Bennett University, and I build things at the intersection of applied AI and full-stack engineering. My work spans from fine-tuning and deploying LLMs to shipping production React/FastAPI applications end-to-end.

Right now I'm a Research Intern at DRDO, working on a computer vision and AI-based situational awareness system. Alongside that, I've built several AI/ML projects from scratch: a hallucination-detection system for LLM outputs (73% F1, benchmarked against multiple baselines on FEVER and TruthfulQA), a financial compliance AI agent that was a finalist at AWS ImpactX (TechFest IIT Bombay, 100+ teams), and a real-time meeting intelligence platform with a FastAPI/WebSockets backend, observability, and integration testing. I also interned as a Software Engineering Intern at Avijo Health, where I shipped 15+ production React components across three live platforms.

I care a lot about rigor — I don't just want models that work, I want to know why they work, where they fail, and how to measure that honestly. That's the throughline in my research too: I'm currently working on inference-time hallucination detection for LLMs at Bennett University.

I'm a fast learner who enjoys picking up whatever a problem demands — Python and ML are my strongest foundation, but I've also built in Java, Vue.js, and Next.js when a project called for it. I'm looking for AI/ML, full-stack, or software development internships/roles where I can keep building real systems, work with people who care about doing things properly, and grow into stronger engineering judgment.

Outside of core project work, I contribute to open source and I'm active in competitive programming (150+ LeetCode problems solved).