Headline

AI Engineer @Axion | Ex@TigerAnalytics | Generative AI & ML Specialist | Global Finalist @Google Solution Challenge | Founder @Heurisko 🇮🇳

About

I build AI systems that actually ship not demos. As an AI Engineer, I work at the intersection of LLMs, agent architecture, and production engineering. I've designed context-aware retrieval systems using graph-based RAG, built custom MCP servers that give AI agents real-time access to live business data, and engineered compliance-grade NLP pipelines that replaced costly vendor tooling all in production environments with real users and real stakes. Before moving fully into AI engineering, I spent 3 years in data science building forecasting models, ETL pipelines, and analytics systems at scale which adds my understanding of data deeply, not just model APIs. What drives me: closing the gap between what LLMs can theoretically do and what they reliably do in production. I care about observability, cost efficiency, and building systems that hold up under pressure. Currently exploring: agentic workflows, multi-model orchestration, and LLM evaluation frameworks. Open to AI Engineer roles where the problems are hard and the bar is high.

Location

Bengaluru

AI stack

claude code

cursor

langchain

langfuse

langgraph

ollama

Work experience

AI Engineer at Axion, Apr 2026 - Present · 4 mo, Bengaluru, ● Built custom MCP server connecting internal DBs and exposing structured tools to a web client, enabling real-time access to client sentiment, deal stages, and real time account status ● Engineered in-house context knowledge graph using GRAG to dynamically surface current account details and relationships, replacing static context injection with graph-aware retrieval ● Built AI Auditor to monitor cost, efficiency, use case coverage, and security issues across all Claude skills, providing org-level visibility into AI spend and performance audit reports weekly

Senior Analyst at Axion, Aug 2025 - Apr 2026 · 8 mo, Bengaluru, ● Developed GenAI agent using LangGraph + OpenAI for end-to-end data preprocessing and validation, integrated Langfuse for observability and Pydantic + Polars for high-performance ● Led Client analytics solutions, guiding 8-member team and taking production decisions

Senior Analyst at Tiger Analytics, Jan 2025 - Jul 2025 · 6 mo, Bengaluru, Karnataka, India, ● Built a sales forecasting solution for Shell’s 1100+ franchise stores using time-series models and feature engineering to improve operational planning and inventory accuracy ● Developed ROI and scenario-planning tools for client Amazon marketing spend across 2,000+ SKUs, and created MMX scripts & Power BI dashboards to track accuracy, detect drift, and guide retraining decisions

DataScience Analyst at Tiger Analytics, Jul 2023 - Dec 2024 · 1 y 5 mo, Chennai, Tamil Nadu, India, ● Designed ETL pipelines for MARS Inc. harmonizing 40M+ sales records across LATAM & APAC, and built Bayesian/heuristic MMX models for KIND Snacks guiding $65M+ annual media investment decisions, also mentored 5+ analysts during AI/ML training as trainer buddy.

Software Development Intern at Miracle Corporation, Jan 2023 - Aug 2023 · 8 mos, Remote, ● Supported cloud solutions team as an intern, specializing in cloud projects. ● Demonstrated AWS proficiency through hands-on tasks in deployment and troubleshooting. ● Collaborated cross-functionally, enhancing cloud infrastructure efficiency and problem-solving.

DSC Lead at Google Developer Student Clubs PSIT, Oct 2022 - Jul 2023 · 10 mos, Kanpur, Uttar Pradesh, India, ● Connected students with Google resources: Facilitated access to Google tools and activities. ● Mentored in tech stacks: Guided students in their tech specialties, fostering skill development. ● Built community ties: Established collaborations with other Google communities, promoting a culture of knowledge-sharing.

Writer at Medium, Jan 2022 - Jan 2023 · 1 yr 1 mo, Remote, I write blogs to share new learnings or interesting ideas in data science, modeling, and AI.

Core Team Member at Google Developer Student Clubs PSIT, Sep 2020 - Sep 2022 · 2 yrs 1 mo, ● Core team member of GDSC PSIT for 2 years, actively contributing to sessions on product development, web design, and organizing hackathons. ● Conducted engaging sessions to impart knowledge and skills, fostering a culture of continuous learning within the community. ● Played a key role in organizing hackathons and speaker sessions, providing valuable opportunities for hands-on experience and networking.

Education

Bachelor of Technology - BTech, Computer Science, PSIT Kanpur (Pranveer Singh Institute of Technology)

Public links

medium: https://medium.com/@agrharsh2000

leetcode: https://leetcode.com/agrharsh2000