Aditya S. profile photo 1 on Tal

Aditya S.

Senior AI/ML Engineer, 8+ yrs shipping LLM, RAG & agentic AI to production. Multi-agent systems with LangGraph, ReAct, MCP. RAG pipelines, DSPy optimization, LoRA fine-tuning, LangFuse observability on GCP Vertex AI.

Bengaluru

8+ years of work experience

Egnyte
Jun 20251 yr 3 mosPresent

India

• Architected and shipped a ReAct agent (LangGraph, Judge0) for sandboxed Python code execution, enabling natural-language Q&A and analytics over spreadsheet data. • Re-architected the agent with a SQL-generation approach on Gemini, reducing inference cost and latency by 10x while improving answer reliability and accuracy. • Led team-wide adoption of DSPy for programmatic prompt optimization, replacing hand-tuned prompts with automated, metric-driven prompt engineering backed by evaluation frameworks. • Led a multi-label NLP classification pipeline (AEC tagger) using PEFT/LoRA fine-tuned SLMs (Gemma, Qwen) on GCP Vertex AI, achieving 95% weighted F1 and 77% exact match accuracy with a 31% improvement in document search quality. • Built SLM classifiers across AEC, FSI and ASK domains, improving query routing and domain accuracy. • Built a deep research agent for automated external data discovery and retrieval, and a ReAct-based agentic orchestration loop for synthetic data generation — boosting classifier accuracy by 40% (relative) and accelerating development cycles.
Nanos - AI Marketing
May 20215 yrs 4 mosPresent

Zurich, Switzerland

• Fine-tuned LLMs (LoRA, QLoRA, GRPO) on Hugging Face / PyTorch for dynamic Google Ads keyword generation, improving CTR and campaign ROAS. • Built and deployed RAG pipelines with vector embeddings for cold-start ad campaigns, grounding generation in historical and external data. • Automated end-to-end campaign creation and monitoring via multi-agent systems (Agno), cutting manual overhead by 40%. • Implemented cross-platform budget optimization using Thompson Sampling and Multi-Armed Bandits, raising campaign efficiency by 20%. • Developed an ad-suggestion engine with transformer models (BERT) and RAG-based LLMs, improving ad relevance by 15%.
Course5i
Jan 20201 yr 5 mosMay 2021

Bangalore

• Built a decision-tree model to predict customer churn and upgrade likelihood for a major retail client, improving retention. • Engineered real-time object detection (YOLO, OpenCV), reducing detection errors by 10%. • Conducted client spend analysis identifying cost-saving opportunities, generating $500K in savings.
OceanFrogs
Sep 20195 mosJan 2020

Greater Bengaluru Area

Standard Chartered Bank
Jul 20181 yr 3 mosSep 2019

Greater Chennai Area

Aditya’s AI Stack

langchain
langchain
langfuse
langfuse
langgraph
langgraph
mcp
mcp
huggingface
huggingface
Agno
Agno

Education

Vellore Institute of Technology

Bachelor of Technology, Information Technology

Vellore Institute of Technology

20144 yrs2018

Join Aditya and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Aditya S. on Tal

Headline

Senior AI/ML Engineer, 8+ yrs shipping LLM, RAG & agentic AI to production. Multi-agent systems with LangGraph, ReAct, MCP. RAG pipelines, DSPy optimization, LoRA fine-tuning, LangFuse observability on GCP Vertex AI.

About

AI ML Engineer specializing in Artificial Intelligence and Generative AI. Actively contributing to projects that leverage advanced agentic workflows to optimize solutions. Holds a BTech in Information Technology from Vellore Institute of Technology.

Location

Bengaluru

AI stack

langchain

langfuse

langgraph

mcp

huggingface

Agno

Work experience

AI ML Engineer at Egnyte, Jun 2025 - Present · 1 yr 3 mos, India, • Architected and shipped a ReAct agent (LangGraph, Judge0) for sandboxed Python code execution, enabling natural-language Q&A and analytics over spreadsheet data. • Re-architected the agent with a SQL-generation approach on Gemini, reducing inference cost and latency by 10x while improving answer reliability and accuracy. • Led team-wide adoption of DSPy for programmatic prompt optimization, replacing hand-tuned prompts with automated, metric-driven prompt engineering backed by evaluation frameworks. • Led a multi-label NLP classification pipeline (AEC tagger) using PEFT/LoRA fine-tuned SLMs (Gemma, Qwen) on GCP Vertex AI, achieving 95% weighted F1 and 77% exact match accuracy with a 31% improvement in document search quality. • Built SLM classifiers across AEC, FSI and ASK domains, improving query routing and domain accuracy. • Built a deep research agent for automated external data discovery and retrieval, and a ReAct-based agentic orchestration loop for synthetic data generation — boosting classifier accuracy by 40% (relative) and accelerating development cycles.

Machine Learning Engineer at Nanos - AI Marketing, May 2021 - Present · 5 yrs 4 mos, Zurich, Switzerland, • Fine-tuned LLMs (LoRA, QLoRA, GRPO) on Hugging Face / PyTorch for dynamic Google Ads keyword generation, improving CTR and campaign ROAS. • Built and deployed RAG pipelines with vector embeddings for cold-start ad campaigns, grounding generation in historical and external data. • Automated end-to-end campaign creation and monitoring via multi-agent systems (Agno), cutting manual overhead by 40%. • Implemented cross-platform budget optimization using Thompson Sampling and Multi-Armed Bandits, raising campaign efficiency by 20%. • Developed an ad-suggestion engine with transformer models (BERT) and RAG-based LLMs, improving ad relevance by 15%.

Data Scientist at Course5i, Jan 2020 - May 2021 · 1 yr 5 mos, Bangalore, • Built a decision-tree model to predict customer churn and upgrade likelihood for a major retail client, improving retention. • Engineered real-time object detection (YOLO, OpenCV), reducing detection errors by 10%. • Conducted client spend analysis identifying cost-saving opportunities, generating $500K in savings.

Data Scientist at OceanFrogs, Sep 2019 - Jan 2020 · 5 mos, Greater Bengaluru Area

Software Developer at Standard Chartered Bank, Jul 2018 - Sep 2019 · 1 yr 3 mos, Greater Chennai Area

Education

Bachelor of Technology, Information Technology, Vellore Institute of Technology