70%
Manual work, removed
Billing and inventory ran on paper and memory. I instrumented a baseline first, automated the highest-cost processes, then measured the delta rather than asserting it. Month-end went from a scramble to audit-ready.
Open to senior analytics, operations & AI enablement roles
I build the systems businesses actually run on.
Turning messy operational reality into measured outcomes — global retail analytics at NielsenIQ, then rebuilding an entire business on automation and applied AI. I don't ship slideware. I ship the thing, then I measure what changed.
Usually replies within a day · Pune, India · Remote-ready across APAC, EMEA & partial US overlap
0%
Manual processing eliminated
0%
On-time delivery at NielsenIQ
0%
Client satisfaction sustained
0
Continents of client delivery
Power BISQLPythonRAG PipelinesOllama
LM StudioMCPLinux / NginxNode.jsSQLite
Process AutomationKPI Governance
Power BISQLPythonRAG PipelinesOllama
LM StudioMCPLinux / NginxNode.jsSQLite
Process AutomationKPI Governance
The work
Three problems.
Three measurable answers.
Every number here has a system behind it that someone still uses today. That is the only standard I hold work to.
100%
Delivery that never slipped
Multi-market retail analytics for international FMCG accounts across North America, Europe and APAC. Concurrent multi-million-dollar contracts, sustained 95% CSAT, and a full cloud migration with zero client escalations.
0
Cloud dependency for private data
Built a retrieval system over internal documents running entirely on local models — a deliberate cost, privacy and latency trade-off against hosted APIs. Staff query SOPs and vendor terms in plain language; nothing leaves the building.
The path
From engineering rigour to operating systems.
Mechanical engineering taught me to model before building. Analytics taught me that the hardest part is agreeing what the number means. Operations taught me that neither matters unless someone actually uses it.
Business Operations Manager
Sujata Fashions · Pune
- Promoted to own business operations end to end — production, inventory, billing, vendor coordination and planning.
- Own vendor negotiation and production scheduling with full P&L visibility and no external team dependency.
- Authored the SOPs, process maps and operational checklists the business now runs on.
- Continue to own and extend the automation and AI systems built in the prior role.
Operations Analyst & AI Systems Lead
Sujata Fashions · Pune
- Designed and deployed automated billing and inventory workflows, cutting manual processing time by 70%.
- Built a RAG pipeline over internal documents with locally hosted LLMs (Ollama, LM Studio) and MCP-connected tooling.
- Built and maintained the business website and operational workflows on a self-managed Linux stack.
- Designed the reporting layer — inventory turns, throughput, vendor performance, margin by line.
International Customer Success & Analytics Analyst
NielsenIQ (NIQ) · Pune
- Owned analytics delivery for international FMCG accounts across North America, Europe and APAC.
- Built executive Power BI dashboards used directly in QBRs by brand and category leadership.
- Defined and governed KPI definitions with clients, ending recurring metric disputes that delayed sign-off.
- Sustained 100% on-time delivery and 95% CSAT on multi-million-dollar annual contracts.
- Managed client communication through a full Google Cloud migration with zero escalations to senior management.
Engineering Internships
Develop Train Maintain LLP · Decibels Lab
- Built and validated mathematical models of mechanical systems in MATLAB/Simulink.
- R&D support for energy-efficient solutions; CAD modelling and technical documentation.
Executive Assistant — concurrent with studies
Sujata Fashions · Pune
- Supported management reporting and inter-departmental coordination while completing a full-time B.E. programme.
Built, not described
Things you can actually open.
Claims are cheap. These are public, and the code is there to read.
Automation · Live system
Billing & inventory automation
The system behind the 70% figure. Scripted validation and reporting over a SQLite store, replacing a manual paper process and eliminating a recurring class of transcription errors.
View repositoryApplied AI · Local-first
Internal retrieval assistant
RAG over SOPs, vendor terms and product specifications, running on local models with MCP-connected tooling. Knowledge that lived in individual heads became queryable in plain language.
Explore GitHubEngineering · Patent-pending
Robotic wheel replacement
Final-year capstone: an automated mechanical handling concept designed to cut manual intervention and turnaround time. Mechanism design, functional reasoning and full technical documentation.
Ask me about itHow I work
Four habits that survive contact with reality.
I use AI tools heavily and I'm precise about where they help and where they don't. That distinction is the job.
01
Baseline before building
A number you didn't measure before the change isn't an outcome, it's a story. I instrument first so the delta is defensible when someone senior pushes back on it.
02
Adoption is the only metric
A system nobody uses is a hobby. I train, document, and iterate against real objections from non-technical staff until the thing is genuinely in daily use.
03
Verify what the model says
I build with AI daily. I also know exactly where it produces confident nonsense — and I design the verification step before I trust the output on anything workflow-critical.
04
Translate in both directions
Most delivery failures are definitional, not technical. Years of sitting between clients, delivery teams and engineering taught me to fix the disagreement before writing the query.
Let's talk
If you're hiring for the gap between data and done — that's my job.
I'm looking for senior analytics, operations, or AI enablement work where ownership is real and the outcome is measurable. If that's the role you're filling, I'd like to hear about it.
B.E. Mechanical · CGPA 9.47/10
Google Data Analytics Certified
Remote-ready, IST based