Headline

Agentic AI @ Google

About

Google forward Deployed Engineer / solution architect with 12+ years of innovation in Data Science and Solution Engineering. I specialize in bridging the gap between customer vision and commercial reality, converting ambiguous business requirements into deployable, high-ROI solutions. By combining deep technical rigor with commercial acumen, I turn Google’s most advanced capabilities—from Multi-Agent systems to Reasoning Engines—into deployable assets. I lead the end-to-end technical lifecycle, from diagnosing complex problems and designing scalable architectures to developing MVPs and leading production rollout, directly accelerating strategic sales cycles for key enterprise accounts.

Location

Greater Bengaluru Area

AI stack

gemini

hugging face

Google ADK

Work experience

AI & GCP at Google, May 2026 - Present · 6 mos

Sr. Solution Architect at Google, Dec 2024 - Present · 1 yr 11 mos, Multi-Agent Architecture Optimization: Re-architected Multi-Agent prompt chains to reduce latency by 50% (60s to <10s) while maintaining 99% accuracy. This technical validation directly unlocked a $100M+ strategic commitment across new business units. Enterprise OCR & Fraud Solution: Engineered and deployed scalable OCR & Fraud extraction solution meeting high-accuracy demands for complex documentation, driving $18M in realized cost savings and establishing a new horizontal standard. Strategic Prototypes & Sales Acceleration: Designed the "AI Skin" blueprint and Tradeflow architecture, establishing Agent-to-Agent (A2A) observability standards. Moved complex processing workstreams from concept to production, solidifying client production roadmaps. Innovation & Knowledge Sharing: Built Gemini-powered Voice Negotiation agents and a centralized Demo Hub to showcase real-time GenAI capabilities, securing C-Level buy-in at major firms. Customer DNA Knowledge Graph: Developed a "Big Bet" integration using BigQuery to map "Customer DNA" for hyper-personalization, now a key asset for validating feasibility across multiple accounts. Thought Leadership: Prototyped an R&D Assistant using LangGraph and Reasoning Engine, compressing research timelines from years to days. Published 4+ technical blogs upskilling external clients and internal engineers.

Principal Engineer - Gen AI at GSK, Jun 2024 - Dec 2024 · 7 mos, Bengaluru, Karnataka, India

Sr. data scientist at Optum, Dec 2022 - Jun 2024 · 1 yr 7 mos, Bengaluru, Karnataka, India, 1. Led a team of Data Scientists to build a Semantic Search for Providers, improving Recall@8 by 20%. This model includes query augmentation using Falcon7b, Vector Search on HNSW, and Ranking using various parameters like ICD9 & ICD10 codes, User Medical Records, etc. 2. Led a team to create an RAG for enterprise policy using Agents & Function calling. 3. Implemented LLMs using quantization & CT2 translate, achieving a 3x improvement in latency. 4. Fine-Tuned the Llama2 model to identify ICD9 & ICD10 codes in EHR data, improving recall by 20% over the benchmark BERT Model. 5. Trained a Multi-Task Language Model for Column Annotation Task, converting enterprise EHR, Claims, & Benefits data to corresponding FHIR representation for interoperability. This model saves 30k+ hours annually in manual data transformation. 6. Developed a chatbot using Embeddings, Semantic Search, & Lang-chain to answer user queries about insurance benefits based on Benefits & Eligibility Contracts and Prompt engineering using React Framework. This improved Recall@8 by 17%.

Principal Data Scientist (NLP) at Aurigo Software Technologies, Dec 2020 - Aug 2023 · 2 yrs 9 mos, India, 1. Implemented an automated Siamese sentence Embeddings and FAISS search for Question-Answering in Request for Proposals (RFPs). This innovative solution significantly reduced the response time from two weeks to just one day, resulting in a remarkable time-saving of over 1000 hours for the sales team. 2. Leveraged KeyWord Phrase extraction (using XLNet Transformers) to identify crucial phrases within RFP documents, enabling the product team to effectively prioritize product features. This approach facilitated the creation of comprehensive product backlogs across multiple teams. 3.Developed and trained a NER (Named Entity Recognition) model using a CRF layer on top of a BILSTM layer, successfully extracting ten entities, including Money, PCT, Organization, and Date, from construction contracts. This model exhibited an impressive F1-Score of 88% across all entities. 4.Spearheaded the engineering efforts for Aurigo Engage, an advanced online feedback solution that employs artificial intelligence to analyze and categorize each comment for public agencies. In this role, I collaborated closely with Product Managers, converting their requirements into actionable engineering backlogs and conducting feasibility studies. Additionally, I designed and implemented the complete data pipeline and microservices architecture for the product, incorporating AWS Kinesis, S3 Buckets, PostgreSQL, DynamoDB, and a Deep Learning Model. Notably, this architecture featured Autoscaling capabilities, enabling the system to handle and process an impressive volume of 13.5K concurrent requests per second in real time. The entire data pipeline was efficiently orchestrated using Airflow. 5.As a Data Science leader, I fostered effective communication and collaboration with cross-functional teams and stakeholders. By thoroughly understanding the business needs, I transformed these requirements into data science problems that could be addressed by my talented team.

Mentor Data Science at upGrad, Oct 2019 - Aug 2023 · 3 yrs 11 mos, As a mentor, I help people from diverse background to transition to data science

Education

Master of Technology - MTech, Data Science & Engineering, Birla Institute of Technology and Science, Pilani