Background · Interests · Where I'm going
About.
Final-year CS student at PES University. First-generation researcher. Building toward a research residency, MTech, and eventually an AGI-focused company — self-funded, deliberately.
Bio
I am a final-year B.Tech Computer Science student at PES University, Bengaluru, and a first-generation researcher. I am currently interning at Nagravision Kudelski, building agentic AI systems that orchestrate across enterprise knowledge bases, data lakes, and communication platforms.
Where I am going: a research lab residency or MTech, then founding an AGI-focused company. Self-funded, deliberately. I believe the most meaningful work comes from taking the long path — building deep technical foundations before trying to scale, and keeping the independence to pursue questions that matter even when they are not fashionable.
I am honest about being early. I do not yet have published papers or a PhD. What I do have is curiosity that does not quit, taste for the right problems, and a willingness to spend years on questions that matter. I would rather build real understanding slowly than collect credentials quickly.
Research interests
Mechanistic Interpretability
Reverse-engineering what neural networks actually compute. The algorithms learned during training, not just the input/output map.
Architecture inductive biases
Attention, state-space, recurrent, convolutional. What each architecture makes easy vs. impossible to learn.
Scaling laws & compute tradeoffs
How compute, data, and parameters trade off, and what that means for choosing between architecture families at a given budget.
LoRA · QLoRA · Quantization
MXFP4, GPTQ, AWQ. Effects on model behavior beyond just parameter efficiency — where the numerics bend what the model can do.
Mesa-optimization
Models that learn to optimize internally during forward passes. Emergent optimization loops that aren't explicitly trained for — the subject of Orpheus.
Mathematical reasoning
Pushing LLMs to solve competition-level mathematics through tool integration and structured reasoning. See Ramanujan.
Education
PES University, Bengaluru
B.Tech, Computer Science and Engineering
Sadhana PU College
Pre-University (12th) · PCMC
SRKVS
SSLC (10th)
Experience
Nagravision Kudelski
Intern — Agentic AI Systems
Building Project Rick — an agentic chatbot that orchestrates across AWS Bedrock, Knowledge Bases, S3, Athena, Glue, and MS Teams. Designing the agent architecture, tool routing, and retrieval pipelines for enterprise knowledge access.
August AI
Intern — Machine Learning
Worked on ML systems for Tata 1mg (grace platform) and Manipal Group. Applied machine learning to healthcare and enterprise use cases.
Competitions
AIMO Prize 3 36/50
Solo entry, 3 weeks of work. Kaggle competition for AI mathematical-olympiad solving. Case study →
Hackverse 2025, MIT Bangalore 7th place
Built WaySafe — a smart tourism-safety application. September 2025.
ARC AGI 3 Active
Currently competing on Kaggle. Abstract reasoning and generalization benchmark.
Contact
Interested in collaborating on research, discussing AI systems, or just want to say hello? I am always happy to connect.