Hi, I'mManjunathan.I build AI products people actually use.
AI Solutions Architect. Builder. Occasional overthinker.
I like working on the part where an AI demo has to become a real product. The messy bits are usually the important ones: retrieval, evaluation, latency, privacy, and the small decisions that make people trust what they are using.
Off-screen, I am usually chasing a badminton slot on 
TurfTown or 
Playo, lifting to a messy gym playlist, or being very normal about Ben 10 through the 
CLI. Online I answer to CodingBad02.
Services
I usually come in when an AI idea has promise, but the path to production is still fuzzy. My job is to make it concrete enough to ship.
Execution
Forward-Deployed AI Systems
I work close to users and operators, find the painful workflow, and turn it into a shipped AI system with measurable adoption.
- +Workflow discovery
- +Prototype to production
- +User-facing rollout
- +KPI instrumentation
Systems
Enterprise AI Architecture
I design RAG, agent, and decision systems around evals, data boundaries, observability, and the parts that make leadership trust them.
- +RAG/agent architecture
- +Evals and guardrails
- +PII-aware flows
- +Observability
Velocity
AI Engineering Acceleration
I help teams adopt the latest coding tools, repos, and AI engineering practices without chasing noisy trends or dead-end abstractions.
- +Agentic dev workflows
- +Repo/tool audits
- +Codegen guardrails
- +What not to build
R&D
Computer Vision & Applied R&D
I turn messy images, video, and sensor data into decision loops: detection, tracking, action understanding, and field-tested feedback.
- +Detection and tracking
- +Action recognition
- +Data engine design
- +Edge/cloud deployment
Products I have built
The part I care about most: turning a useful idea into something with a name, a logo, and people on the other side of it.
Budhi AI
A memory layer for notes, files, and moments you do not want to lose.
1000+ downloads
Planr AI
Turns messy ERP/WMS data into shortage signals, ranked actions, and what-if planning.
Manufacturing AI
V-Commerce Studio
A hackathon build for agent-led shopping flows and product discovery.
GKE hackathonBy the numbers
A few operating numbers from the work: users reached, retrieval quality, enterprise constraints, and deployment speed.
See the workFortune 500 clients
Enterprise AI systems shipped
Retrieval quality
Across applied AI projects
Budhi AI downloads
Second-memory users reached
Faster deployment time
Kubernetes release speed-up
Contact
If you are building something useful with AI and want a second brain on the hard parts, write to me.