CS × AI/ML AT IIITS
Sai Teja Ala
Hi! I like ML research, foundation models, and AI for scientific discovery. I’ve worked on problems spanning foundation models, clinical AI, and computational neuroscience. Long term, I want to build intelligent systems that can reason, discover, and help accelerate scientific progress.
Research / Projects
MedGuard: KG-Grounded Selective Prediction
NEURIPS 2026 (UNDER REVIEW)First author. Identified the Numerical Cliff failure mode in multimodal LMs and developed a knowledge-grounded selective prediction framework combining UMLS knowledge graphs with conformal calibration.
02Caltech Prober Lab — Neural Population Analysis
CALTECH (REMOTE)Developed ML & signal-processing pipelines for high-dimensional calcium imaging (>10K neurons), low-rank embeddings, and automated HCRT neuron identification under pharmacological perturbation.
03Reliable Multimodal Reasoning with Structured Knowledge
IIIT SRI CITY HONORS THESISInvestigating multimodal LLM reasoning under conflicting evidence using GraphRAG, uncertainty calibration, document-parsing front-ends, and structured report generation.
04Model-Free Reinforcement Learning from Scratch
REINFORCEMENT LEARNINGImplemented 6 model-free RL algorithms (Monte Carlo, TD(0), TD(λ), GLIE, SARSA, SARSA(λ)) analyzing convergence, exploration, and credit assignment across a 400-state environment.