Headline

Senior Software Engineer | Applied AI, LLMs, RAG & Agents | Production-Scale Systems (1M+ req/hr) | Relyance AI | IIMB 🎓

About

I’m a Senior Software Engineer with 8+ years of experience building distributed systems, cloud infrastructure, and production applications — increasingly focused on applying AI to real-world products. My recent work spans LLMs, RAG, vision-language retrieval, audio intelligence, and agentic workflows. I’ve worked with ColPali and Qdrant for vision-language retrieval, built multilingual audio pipelines using Whisper + LLMs + KeyBERT with Airflow, and explored production agent workflows with LangGraph. My systems background is what I bring to AI: designing for scale, reliability, latency, and cost. I've worked on systems handling 1M+ requests/hour at 30ms p95, multi-region DynamoDB Global Tables, Lambda@Edge, and Terraform. I've delivered 100x latency reductions on critical paths and driven ₹2Cr+ in infrastructure savings through automation. I'm particularly interested in the engineering challenges that emerge when AI moves beyond prototypes — evaluation, retrieval quality, latency, cost, observability, and reliability. Currently going deeper into grounded RAG, agentic systems, AI evaluation, and production LLM infrastructure. Open to Applied AI / AI Product Engineer / Senior Software Engineer roles — Bangalore or remote.

Location

Bengaluru

AI stack

claude code

hugging face

langchain

langgraph

langsmith

ollama

Work experience

Senior Software Engineer II at Relyance AI, Feb 2025 - Present · 1 yr 8 mos, Bengaluru, • Architected a multi-region, edge-routed consent platform (30ms p95 latency, 1M+ requests/hour and 100K+ records/days, 99.9%+ availability) using CDN + serverless (Lambda@Edge) + distributed storage (dynamoDB Global Tables). • Delivered authenticated cross-device consent sync (user-level), 100+ subdomain propagation support per domain, and global compliance (GPC, WCAG, 60+ locales) enabling consistent enforcement at scale. • Built a web consent agent with dynamic + pattern-based blocking, preventing 100% of tracking injection paths (scripts, iframes, pixels, beacons) with near-zero runtime overhead (<5ms impact) & zero-code integration to ensure enterprise-grade privacy compliance. • Automated Infrastructure via Terraform, streamlining the deployment of event-driven analytics pipelines (DynamoDB Streams → Lambda → Pub/sub → BigQuery → MongoDB).

Senior Software Engineer at Bonito Designs, Apr 2023 - Feb 2025 · 1 yr 11 mos, Bengaluru, ML Audio Intelligence Pipeline (Python, Airflow, OpenAI, Whisper, KeyBERT) enabling transcription, translation, summarisation & keyword extraction, delivering multi-language conversational intelligence & insights. Built a vision-language semantic search system from scratch using Colpali patch embeddings, Qdrant for multi-vector late-interaction retrieval over 300+ vendor PDFs – search time minutes to 2 seconds, saving 500+ operational hours monthly. Bonito CustomerApp (Flutter GetX, Firebase, NodeJS, Django REST, Go, MySQL, MongoDB, AWS) • Architected CMS-driven, plug-and-play frontend for Android, iOS, and Web with Firebase & Google Analytics integration • Designed micro service architecture (Project, User, Notification, CMS) with CDN, API Gateway & CI/CD on EKS • Scaled platform to support 1k+ concurrent users with ~350ms latency • Owned full SDLC including architecture design (HLD/LLD), deployments, releases, reviews & technical documentation SketchUp Automation – 3D to 2D & Estimation (Ruby, React, NodeJS) • Automated 2D drawing generation using SketchUp Ruby APIs, reducing manual effort by 90% with 99% dimensional accuracy (≤1mm); copyright filed • Built scalable 3D tagging system, cutting 2-day development effort to 5 minutes • Automated 15+ estimation categories from 3D models using complex algorithms and UOM conversion dashboards • Achieved 90%+ first-time-right, zero NCs, eliminated designer roles, saving ₹2 Cr • Reduced 2D design cycle time from 2 days to 30 minutes, eliminating rework loops Mentorship & Performance Optimization • Mentored engineers on automation systems, Redux/Saga, and Flutter fundamentals • Reduced API latency from 45s to 200ms via precomputation strategies • Optimized file uploads using pre-signed S3, cutting latency from 40s to 3s per MB

Software Development Engineer - III at HomeLane, Apr 2022 - Mar 2023 · 1 yr, Bengaluru, Karnataka, India, HomeLane CustomerApp (NodeJS/Django Rest/Mysql/Mongo/Flutter/AWS) - Led a team of 5 members in developing and implementing system architecture designs, patterns, and approaches led to efficient quality secure app by conducting quality code reviews, deployments, releases, and documentation. - Created comprehensive project plans, including documentation, flows, timelines, logical execution order, and ensured software quality and on-time delivery. - Developed Flutter app using design patterns following Play/App Store guidelines using GetX/Redux frame-work and integrated with Firebase and GA.

Software Development Engineer - II at HomeLane, Oct 2020 - Mar 2022 · 1 yr 6 mos, Bengaluru, Karnataka, India, Tech Stack Migration - (PHP to DjangoRest & React/Redux) - Optimized 60+ API response time and secured 100+ public APIs and build scalable architecture through stack migration. - Developed Django Rest APIs with authentication and authorization, using Swagger and JWT tokens. Order Management System (Django REST/Mysql/ReactJs/NodeJS) - Delivered streamlined order cancellation and refund flow, improving customer experience. - Designed, and delivered Order Management System, including migrating 5K+ orders data into the system, resulting in a streamlined process. - Implemented web-hook and inter/intra-communication with Salesforce, ERP, Acuity, and other internal systems to synchronize data between multiple systems.

Software Engineer at HomeLane, Apr 2019 - Sep 2020 · 1 yr 6 mos, Bengaluru, Karnataka, India, Franchise Commission Management System (PHP/Mysql) - Delivered Franchise Commission Management System before schedule, resulting in 80% manual effort reduction for finance team commission calculation and tracking. Automated email/notification triggers introducing database templating, reducing development efforts from 1 day to 5mins.

Software Engineer at Simplilearn, Oct 2018 - Mar 2019 · 6 mos, Bengaluru Area, India, Worked on Back-end of Payment module, Cost Tracking module, Course Extension Module, Exam voucher Module. Develop, unit-test, maintain design documentation of code for release process.

Intern at Simplilearn, Mar 2018 - Sep 2018 · 7 mos, Bengaluru, Karnataka, India

Education

Two-Credit Course, New Product Development, Indian Institute of Management Bangalore

PG-DAC, Pg Diploma in advance computing, Centre for Development of Advanced Computing (C-DAC)

Public links

github: https://github.com/AkashAgarwalInd