May 2025 – Jul 2025 · 2 mos · Bangalore Urban, Karnataka, India · On-site

• Integrated real-time voice-driven bidirectional streaming APIs into AWS Q for Business using Coral (Amazon’s Java/Spring framework), enabling seamless voice-to-LLM-voice interactions via Bedrock-hosted models and EC2-deployed services.
• Architected and deployed a scalable microservice-based voice interface as an independent service layer within AWS Q, using Amazon Bedrock, SageMaker, and EC2, and designed with PACELC to ensure persistent sessions with low-latency.

Apr 2025 – Jun 2025 · 2 mos · Greater London, England, United Kingdom · Remote

• Engineered multi‑layered SQL validation (basic & schema‑aware) to block dangerous operations and mis‑typed tables/columns, reducing invalid‑query failures by 85%.
• Developed a complete text->sql multi agentic system; scaled to handle 100K+ concurrent requests; rule‑based fixes for common typos, LLM‑based fallback for complex errors, a reranking; self‑learning loop—that auto‑corrects 92% of syntax/name mistakes and prevents 60% of recurring error patterns.

Mar 2025 – Apr 2025 · 1 mo · San Francisco Bay Area · Remote

• Resolved support tickets via Notion, Linear, and Slack; on user onboarding by deploying seamless OCR and parameter extraction pipelines from submitted legal documents, improving user intake efficiency by 23%.
• Engineered a multi-agent document extraction pipeline, increasing OCR accuracy by 45% and reducing data-entry errors, using: Mistral-ocr-latest for markdown conversion; Pixtral-12B and Mistral-2B SLM as a post-processing agents.

Dec 2024 – Feb 2025 · 2 mos · Boston, Massachusetts, United States · Remote

• Building a biomedical knowledge graph pipeline processing 500+ PubMed abstracts with MedPalm-2 and GPT-3.5, achieving 92% accuracy in protein-pathway mapping cross-referenced with STRING-db, and Reactome.
• Scaled automated graph construction by 22.71%, creating 3000+ entities and relationships. used multi-agent ar- chitecture for, leveraging small language models (SLMs) to reduce inference time and latency, achieving a 14.16% improvement in query response time.

May 2024 – Jul 2024 · 2 mos · Boston, Massachusetts, United States · Remote

• Underguidance of Prof.Dong Xu and Dr.Gyan Shrivastava from University of Missouri ; proposed an architecture combining BioBERT + LoRA and DistilBERT with 97% text understanding, facilitating drug target discovery.
• Fine-tuned BioBERT with Low-Rank Adaptation (LoRA), reducing parameters by 90% while maintaining 82.4% performance and achieved a 11.275% improvement in biomedical information retrieval.
• Implemented TechGPT-2 for knowledge graph, fine-tuning on QLoRA weights, achieving 90% NER, RTE accuracy

Jan 2024 – May 2024 · 4 mos · Mumbai Metropolitan Region · Remote

• Building mathematical models to predict market variations(alphas) across various developed and emerging sub-industry markets using price-volume, company fundamental, and sentimental datasets.
• Backtested novel and improved strategies over the last 10 years of data on WorldQuant's software using FastExpression language to enhance the latency and compare the graphs of each individual tested strategies.
• Collaborating with reading research papers on alpha creation and generating stochastic processes to find optimum strategies or loopholes in the volume/equity markets to generate alphas.

Sep 2023 – May 2025 · 1 yr 8 mos · Prayagraj, Uttar Pradesh, India

Sep 2023 – Apr 2024 · 7 mos · Prayagraj, Uttar Pradesh, India · On-site