Hello, I'm
Poovarasu Sekar
AI engineer and Team Lead with 7+ years building production software. I've grown from full-stack into agentic AI — turning LLMs into dependable products (RAG, autonomous agents, on-prem LLMs, GenBI) — while leading and mentoring the Python and AI Engineering teams that ship them.
role="AI Architect",
llm="anthropic/claude-opus",
experience="7+ yrs",
traits=[
"Spec Driven Development",
"Autonomous Agents"
]
)
About
About Me
I build software that's a joy to use — and AI you can actually trust.
I'm a Team Lead and AI engineer with 7+ years at one company — six promotions from intern to lead. I started in PHP/Laravel full-stack, went deep in Python, Django and AWS, and today I focus on AI Engineering: production RAG systems, autonomous agents, and Generative BI. A Mechanical Engineering grad turned self-taught software engineer, I care about craftsmanship, clarity, and shipping AI that's measured, observable, and genuinely reliable.
- Role
- Team Lead — Python & AI
- Expertise
- AI & Python Engineer
- Current Focus
- AI Engg. — Agents, RAG, GenBI
- Experience
- 7+ Years
- Mobile
- +91 8148406208
- contact@poovarasu.dev
- Education
- B.E. Mech. — Kongu Engg. College
- Location
- Coimbatore, Tamil Nadu, India
- Production AI — RAG & agents
- End-to-end product delivery
- Eval-driven & observable
- Team leadership & mentoring
- Founded two teams from scratch
- Research first, then design, then code
"Build systems that remain understandable, scalable, and valuable long after the first version is shipped."
Impact
By the Numbers
Measured outcomes from the platforms, teams and AI systems I've built and led.
What I do
My Expertise
A snapshot of the stacks I lean on every day to ship resilient, considered software end-to-end. Click any card to dive deeper.
Career
My Work Experience
From PHP intern to AI-focused team lead — 7+ years of shipping production software, growing through full-stack, Python, AWS, data, and now AI engineering & leadership.
Six promotions, one company — still growing here
Team Lead
Mallow Technologies
Heading both the Python and AI Engineering teams — two teams I started from scratch. I keep driving AI-first delivery while owning team leadership, architecture, engineering standards and mentoring across both groups.
- Leading the Python (~20+ engineers) and AI Engineering (~10 engineers) teams — delivery, architecture and engineering standards across both
- Managing 5+ projects in parallel as technical lead and project owner; running code reviews, mentoring engineers and acting as the primary client contact
- Maintaining a 3TB+ production database with 5+ replicas at up to 40K requests/minute; resolved a critical RDS misconfiguration incident with zero application code changes
Selected work
Key Projects
Platforms I've built, led and brought AI into. Client names are kept confidential — the work speaks for itself.
Taxi Booking Platform
Ride-hailing, like Ola / Uber
My flagship project and the thread through my whole career — I joined it as a developer trainee, grew into leading its backend team, moved it onto the Python team, picked up its analytics track, and today lead its AI & Analytics team.
- Multi-agent complaints system that investigates, acts, resolves and initiates refunds autonomously — daily complaints down from 500+ to under 50
- Demand forecasting & hotspot intelligence by zone and hour (weather, events, flights/trains, festivals) driving driver-positioning heatmaps
- Competitor fare-monitoring agent, competitive-intelligence feed and a call-recording analysis agent that rates call-centre quality
- Admin GenBI chatbot (WrenAI) for plain-English questions over the full dataset, plus context-aware, behaviour-triggered notifications
- Self-hosted OSRM replacing paid routing APIs (30% / $29.7K a year saved) and serverless Lambda reporting at ~$2/month
- Scale: 72 lakh users, 20K drivers, 13+ cities, 100 → 60K bookings/day, 5.5 crore total bookings, 30K–40K requests/minute, 14 modules
Marketplace Seller Analytics Product
Seller portal for Amazon / Walmart sellers
A two-era project: I built PHP features on it as a junior developer, then returned to lead its AI Engineering — growing from basic GenAI features into complex multi-agent systems.
- Public-facing MCP server with detailed, domain-specific skills across multiple use cases
- Built-in multi-agent chatbot that reads and writes across domains and reasons over many data points — tied to revenue and sales outcomes
- GenAI embedded directly into core product workflows
- Earlier era: product information management, feedback & review automation and a reimbursement management system
Group Buying Platform
Multi-tenant community commerce
Connects manufacturers, vendors and distributors with group members for group buying and community management — started as senior developer, now leading 5+ developers & QA.
- AI chatbot built as a multi-agent system — domain-specific agents that together operate the whole application through natural language
- Autonomous multi-agent system for anomaly detection and management
- AI file-transformation engine that restructures large Excel files against an import template
- Group management, group buying, events, announcements and file sharing on a schema-per-tenant architecture
Approach
How I Work
I rarely accept the first solution. I research, compare and design before I implement — because long-term maintainability beats short-term speed.
Step 01
Understand
Start from the business problem and its constraints, not the tech.
Step 02
Research
Look at how the industry and larger companies solve it, and what it costs.
Step 03
Compare
Put multiple approaches and technologies side by side.
Step 04
Weigh trade-offs
Scalability, maintainability, cost, flexibility, DX and operational load.
Step 05
Design
Architect in layers — API, services, workers, queues, data, analytics, monitoring.
Step 06
Build & measure
Prototype, validate the assumptions, optimise, then document and present.
What I value
- Maintainability over shortcuts
- Scalability over convenience
- Automation over repetition
- Documentation over tribal knowledge
- Evidence over opinions
- Production readiness, always
How I lead
- Enable developers rather than just assign tasks — every code review is a teaching moment
- Resource and project planning, architecture reviews and sprint planning
- Training sessions, interview prep and knowledge-sharing documentation
- Bridging business and engineering — requirements, presentations and client communication
Tools of the trade
Skillset & Expertise
The technologies I reach for, grouped by area. Pick a category to filter the stack.