Senior Software Engineer · Best Buy India · Bengaluru, India
Anshul Akotkar
Frontend-heavy full-stack engineer building AI-powered experiences at scale.
I build web and mobile products that millions of people rely on. At Best Buy I lead the post-purchase Service Triage platform — an LLM-powered, agentic support experience that serves 8M+ customers a year across web, app, chatbot, IVR and associate tools.
- customer.intent“My laptop won’t charge” + order history
- orchestrator.agentGemini on Vertex AI — plans & routes
- ├─ vision.subagentphoto · video · live camera
- ├─ voice.subagentIVR · speech
- └─ mcp.toolsorders · warranty · scheduling
- → resolutionIn-home repair appointment booked
- Revenue contribution thanks to Service Triage
- Customers a year across 5 touchpoints
- Requests handled at holiday peak
- Fewer abandoned service requests (50% → 30%)
About me
9 years of shipping products people use
- React
- Next.js
- React Native
- Node.js
I've spent 9 years building products with React, Next.js, React Native and Node.js. My sweet spot is the frontend, but I own features end to end — from API design and middleware to CI/CD pipelines, deployments and production monitoring.
At Best Buy India I work on the Self Service Rapport team. We replaced a rigid, menu-driven support flow with a context-aware experience grounded in each customer's purchase history and intent. It routes people to the right option — remote support, chat, an in-store or in-home appointment, or a repair through our insurance partner — and handles close to a billion requests at holiday peak.
Lately I've been going deep on agentic AI: an orchestrator agent with sub-agents that understand text, photos, video, live camera and voice, built on Gemini via Vertex AI, plus MCP clients and servers that connect our agents to internal services and let other teams reuse them.
I also enjoy the people side of engineering — mentoring engineers, leading the team, and writing the docs and processes that make everyone faster.
Off the clock
When I'm not shipping code, I'm probably planning the next trip ✈️, hunting down a cuisine I haven't tried yet, or unwinding with music. Total softie for dogs and cats 🐶, and every other pet too (except rats).
Cricket
ODIs are appointment viewing when Ro-Ko bat
F1
Lewis is the GOAT, and I love watching Max
Tennis
Still missing the Roger–Rafa–Djoker era
…and pretty much any other sport, watching or playing.
Experience
Where I’ve built things
Sep 2024 — Present
2 yrs 3 mos
Senior Software Engineer
Best Buy India · Bengaluru
Lead the LLM-powered post-purchase support platform serving 8M+ customers a year, from React / React Native / Next.js front ends to Node.js services. Built a multi-agent, multimodal AI system on Gemini via Vertex AI, with an MCP layer now reused by other teams.
- React
- React Native
- Next.js
- Node.js
- Gemini · Vertex AI
- MCP
- GCP
- AWS EKS
Nov 2020 — Sep 2024
3 yrs 11 mos
Senior Software Engineer
MFine · Bengaluru
Built MFine’s healthcare app and web platform (1M+ downloads) for doctor consults over chat, audio and video. Led a 12-person team on Corporate Wallet: 60+ corporates onboarded, +47% daily active users, +$500K ARR.
- React Native
- React
- Node.js
- Payments (UPI)
- SSL pinning
Jun 2017 — Nov 2020
3 yrs 6 mos
Product Engineer
HED Experts · Bengaluru
Built the Xcelerator skills platform from scratch: React web app and Elixir/Phoenix services. Designed a microservices architecture that cut deployment time by 70%, and SEO work that lifted organic traffic by 15%.
- React
- Elixir
- Phoenix
- Microservices
- SEO
Jan 2017 — May 2017
5 mos
Web Development Trainee
Publicis Sapient · Noida
Built a trading app with real-time stock market data and D3 visualisations.
- AngularJS
- Node.js
- PostgreSQL
- D3
Education
B.Tech, Computer Science and Engineering
Vellore Institute of Technology
Jun 2013 — May 2017
Selected work
Impact I’ve shipped
Production systems, not side projects — built for millions of customers and the teams behind them.
Flagship · Best Buy India · Lead engineer
Service Triage platform
An LLM-powered, agentic post-purchase support experience. It replaced a rigid menu flow with context-aware routing — remote support, chat, in-store or in-home appointments, or repair through our insurance partner.
- $20–25M
- saved / year
- 8M+
- customers / year
- 5
- touchpoints
- Next.js
- React Native
- Node.js
- Gemini · Vertex AI
- Web · App · Chatbot · IVR
Agentic AI
Multimodal agents + MCP
Orchestrator with sub-agents for text, photos, video, live camera and voice. MCP servers expose internal services to other teams.
- MCP
- Sub-agents
Scale
Holiday-peak ready
Close to a billion requests at peak, with service abandonment down from 50% to 30%.
Delivery
Commit → production
Multi-environment pipelines across OpenShift, AWS EKS and GCP.
- GitHub Actions
- OpenShift
- AWS EKS
Observability
Knowing what prod is doing
Dashboards and alerts on failure rates, plus mobile session tracking.
- Kibana
- Dynatrace
Leadership
Making the team better
Mentoring engineers, leading the team, and writing the architecture docs and processes.
- Mentoring
- Architecture docs
Core skills
What I work with
Frontend
Fast, accessible interfaces for web and mobile.
- React
- Next.js
- React Native
- TypeScript
- JavaScript (ES2023+)
- HTML & CSS
- Web performance
Backend & APIs
Services and contracts the frontend can trust.
- Node.js
- REST APIs
- GraphQL
- Middleware
- API design
AI & Agents
LLM features that ship to production.
- Agentic workflows
- Orchestrator / sub-agent design
- Model Context Protocol (MCP)
- Gemini on Vertex AI
- Multimodal input (text, image, video, voice)
- Human-in-the-loop guardrails
Cloud & Delivery
From commit to production, repeatably.
- GitHub Actions
- Artifactory
- OpenShift
- AWS EKS
- Google Cloud (GCP)
- Multi-environment deployments
Observability
Knowing what production is doing.
- Kibana dashboards
- Dynatrace
- Alerting on failure rates
- Mobile session tracking
Leadership
Making the team better, not just the code.
- Technical leadership
- Mentoring
- Architecture docs
- Engineering process
Let's talk
Let's build something people rely on.
Open to conversations about frontend platforms, agentic AI and engineering leadership.