Akarsh Gajbhiye profile photo 1 on Tal

Akarsh Gajbhiye

Founder | Reliability tooling for AI agents

Bengaluru

4+ years of work experience

Apr 20265 mosPresent

Bengaluru

Reliability tooling for AI agents
Freelance
Mar 20266 mosPresent

Bengaluru

Senior AI + backend execution for startups — agentic systems, RAG, infra, the hard parts of shipping → Building Kairos: Open source Agent Failure Clustering and Evals → Technical lead for Leadzo : built AI driven lead generation and outbound campagins, cloud infra, deploy automation, anti-fraud systems; led technical hiring and architecture → Freelance projects for multiple clients from US → Open to select freelance & consulting
Coreworks AI
Oct 20257 mosApr 2026

Architecting the 0→1 foundational infrastructure and agentic platform layer for an enterprise Document AI system. - Multi-Tier Execution: Re-engineered the core ingestion platform into a multi-tier system, dynamically routing workloads serverless async workers based on queue pressure . - Agentic Systems: Designed and productionized an agentic document understanding system using LangGraph, orchestrating multi-step extraction, validation, and autonomous correction workflows. - High-Scale Ingestion: Built distributed async enrichment engines for massive Excel workbooks (~500K rows) using presigned S3 uploads and Redis-backed workers. - Performance Engineering: Resolved production OOM and latency failures by introducing early file-shape detection, incremental chunking, and memory-stage isolation. - Scaling Infra and Platform for AI Native Platform .
Blitz
Jun 20223 yrs 5 mosOct 2025

Bengaluru, Karnataka, India

Primary owner of the core dispatch and logistics platform, architecting and scaling systems to handle 1M+ daily requests across distributed services. - Quick Commerce & Dispatch Verticle: Led the development of the high-concurrency dispatch engine, managing real-time assignment algorithms for rapid fulfillment. - Real-Time Order Rollout & Guaranteed Assignment: Engineered a high-throughput offer-acceptance system handling intense race conditions. The system ranks available riders based on proximity and SLA constraints using Google OR-Tools, rolls out offers, and ensures atomic, guaranteed assignment to the first responder while instantly neutralizing competing offers. - Geospatial RAG & Intelligence: Built a production RAG system that ingested real-time delivery coordinates and merged them with unstructured data to create a high-precision address intelligence layer. - Applied LLMs: Fine-tuned and self-hosted LLaMA 3.1 models for address normalization and NER, deploying GPU-backed inference on RunPod into production geolocation workflows. - High-Scale Architecture: Built the core backend (Java/Python) for order lifecycle and inventory, optimizing async execution and query patterns to maintain high-write throughput. - Technical Leadership: Led execution for the platform team and mentored two engineers in building production-grade distributed systems.
Poli
Feb 20225 mosJun 2022

Built the "Control Room" for community managers to measure and automate engagement across Slack and Discord. Platform Engineering: Developed the core analytics engine that ingested and processed real-time activity data from Slack and Discord APIs to track community health and growth metrics. Zero-to-One Execution: Owned the full-stack development from ideation to initial user onboarding, participating in high-level venture capital meetings and product-market fit iterations. Infrastructure: Architected a scalable data pipeline to handle event-driven messaging data, ensuring low-latency reporting for active communities.

Education

Indian Institute of Technology, Kanpur

Bachelor of Technology - BTech, Computer Science

Indian Institute of Technology, Kanpur

20155 yrs2020

Join Akarsh and 10k+ other professionals on

Tal Boss

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

Download Tal Boss

Akarsh Gajbhiye on Tal

Headline

Founder | Reliability tooling for AI agents

About

I build AI systems that survive contact with production. What I do → Agentic systems & orchestration → RAG & document-AI pipelines → The backends, integrations & infra that keep them running Where I go deep → Agent reliability & AI security → Adversarial testing, failure-mode analysis → The hard question: does the agent actually do its job once it's out of the demo? Background → Founding engineer : Document AI Agents → Years on large scale backends: geo-intelligence, event-driven pipelines, millions of requests/day Currently building my own thing in AI space . Open to select freelance & consulting alongside it. DMs open.

Location

Bengaluru

Work experience

Founder at To Be Announced, Apr 2026 - Present · 5 mos, Bengaluru, Reliability tooling for AI agents

Independent AI Engineer - Agent Reliability & Audit at Freelance, Mar 2026 - Present · 6 mos, Bengaluru, Senior AI + backend execution for startups — agentic systems, RAG, infra, the hard parts of shipping → Building Kairos: Open source Agent Failure Clustering and Evals → Technical lead for Leadzo : built AI driven lead generation and outbound campagins, cloud infra, deploy automation, anti-fraud systems; led technical hiring and architecture → Freelance projects for multiple clients from US → Open to select freelance & consulting

Founding Engineer at Coreworks AI, Oct 2025 - Apr 2026 · 7 mos, Architecting the 0→1 foundational infrastructure and agentic platform layer for an enterprise Document AI system. - Multi-Tier Execution: Re-engineered the core ingestion platform into a multi-tier system, dynamically routing workloads serverless async workers based on queue pressure . - Agentic Systems: Designed and productionized an agentic document understanding system using LangGraph, orchestrating multi-step extraction, validation, and autonomous correction workflows. - High-Scale Ingestion: Built distributed async enrichment engines for massive Excel workbooks (~500K rows) using presigned S3 uploads and Redis-backed workers. - Performance Engineering: Resolved production OOM and latency failures by introducing early file-shape detection, incremental chunking, and memory-stage isolation. - Scaling Infra and Platform for AI Native Platform .

Sr. Software Development Engineer at Blitz, Jun 2022 - Oct 2025 · 3 yrs 5 mos, Bengaluru, Karnataka, India, Primary owner of the core dispatch and logistics platform, architecting and scaling systems to handle 1M+ daily requests across distributed services. - Quick Commerce & Dispatch Verticle: Led the development of the high-concurrency dispatch engine, managing real-time assignment algorithms for rapid fulfillment. - Real-Time Order Rollout & Guaranteed Assignment: Engineered a high-throughput offer-acceptance system handling intense race conditions. The system ranks available riders based on proximity and SLA constraints using Google OR-Tools, rolls out offers, and ensures atomic, guaranteed assignment to the first responder while instantly neutralizing competing offers. - Geospatial RAG & Intelligence: Built a production RAG system that ingested real-time delivery coordinates and merged them with unstructured data to create a high-precision address intelligence layer. - Applied LLMs: Fine-tuned and self-hosted LLaMA 3.1 models for address normalization and NER, deploying GPU-backed inference on RunPod into production geolocation workflows. - High-Scale Architecture: Built the core backend (Java/Python) for order lifecycle and inventory, optimizing async execution and query patterns to maintain high-write throughput. - Technical Leadership: Led execution for the platform team and mentored two engineers in building production-grade distributed systems.

Co-Founder at Poli, Feb 2022 - Jun 2022 · 5 mos, Built the "Control Room" for community managers to measure and automate engagement across Slack and Discord. Platform Engineering: Developed the core analytics engine that ingested and processed real-time activity data from Slack and Discord APIs to track community health and growth metrics. Zero-to-One Execution: Owned the full-stack development from ideation to initial user onboarding, participating in high-level venture capital meetings and product-market fit iterations. Infrastructure: Architected a scalable data pipeline to handle event-driven messaging data, ensuring low-latency reporting for active communities.

Education

Bachelor of Technology - BTech, Computer Science, Indian Institute of Technology, Kanpur