Feb 2022 – Apr 2025 · 3 yrs 2 mos · Hyderabad, Telangana, India · Hybrid

• Implementing novel interpretability metrics to evaluate LLM decision-making processes
• Clinical Diagnosis Engine development (85.83% accuracy) using Graph RAG
• Multi-lingual contextualized medical entity search team, improved over the trie-solution with appropriate ML techniques keeping the latency in mind
• Fine-tuned 15+ LLMs, including Dr.Samantha-8B (5000+ downloads) and SynthIQ-7B (20,000+ downloads)
• Enhancing long-text coherence in LLMs by integrating episodic memory mechanisms

Oct 2018 – Feb 2022 · 3 yrs 4 mos · Bengaluru Area, India

• Contributed to the development of EazyML, an AutoML platform with explainability for tabular data (as a core R&D engineer)
• Built a skin disease classification system (F1 score: 0.64 across 65 diseases)
• Engineered a state-of-the-art zero-shot concept matching model using BERT just within days of its launch
• Refactored legacy codebases using modern Python OOP practices

Jul 2018 – Sep 2018 · 2 mos · Bengaluru, Karnataka

• Conducted research on Sketch-GAN under the Video Analytics Lab, optimizing CUDA functions to achieve a 1.2x speedup in model performance, enhancing computational efficiency for generative adversarial networks.
• Designed and developed an user-friendly PyQt application to streamline data annotation workflows, improving productivity and reducing manual effort for research teams.

Jul 2017 – Jun 2018 · 11 mos · Goa

• Researched the intersection of Game Theory and Machine Learning, implemented optimized Banzhaf Decision Forests to translate abstract theoretical concepts into functional code.
• Investigated the applicability of captioning models for Visual Image Search, developed a solution using ElasticSearch and TensorFlow. The associated GitHub repository garnered 300+ stars, demonstrating its impact and adoption.
• Published a research paper in SocPros 2017, which has received 19 citations as of 2025