Six ways to work together, from a one-report diagnostic to a system built and shipped.
One system, one written report. What is going wrong, where the choke points and bottlenecks are, and the fixes I would make, in one to two weeks.
you get ✦ a report you can hand to your engineers
I model the unit economics layer by layer, then the stack gets chosen against that model rather than against a benchmark. You see the per-unit cost of every layer before anyone commits to it.
you get ✦ a per-unit projection you can check against real spend
I build the system, put it in production, and stay long enough to see what it does once real users arrive. Cost and behaviour are instrumented from the first commit, not added later.
you get ✦ a system in production with its numbers visible
For teams with their own engineers who want a second opinion on architecture, cost and evaluation choices before they commit to them.
you get ✦ a standing review on decisions as they come up
Whether a system can be explained, audited and defended before it ships. Grounded in the same framework as my C-RAI policy draft and the responsible AI handbook.
you get ✦ findings against a named framework, not a vibe check
Sessions on production AI, cost modelling and evaluation, run as workshops for your engineers rather than as lectures.
you get ✦ your engineers able to do it without me next time
The method does not change much between sectors. What changes is the data, the regulator and what counts as a wrong answer. These are the ones I have shipped in.
Voice agents that take bookings and handle support calls in Hinglish, around the clock. I diagnose them from real call recordings and hand back a labelled failure taxonomy with per-category rates, not an architecture diagram.
seen in ✦ narrowband call audio, barge-in limits, stall signatures
Agents that detect the caller's language, answer from your own knowledge base rather than the open internet, and escalate to a human the moment retrieval comes back thin.
seen in ✦ sub-second response budgets, graceful fallback on empty retrieval
Natural language to SQL over production schemas, with intent routing and semantic table selection so only the relevant tables ever reach the model. Costed per layer before a line of code is written.
seen in ✦ facilities management SaaS, Azure SQL, three-tier cache
Distress and failure signals, trade alerting, and transaction anomaly detection. Every prediction carries an attribution, so an analyst can see which inputs drove a flag before acting on it.
seen in ✦ ESOP workflows, live alert dashboards, XGBoost with SHAP
Clinical assistants that read a provider's curated material instead of general model knowledge, because general models hallucinate confidently on medical advice.
seen in ✦ PDF and spreadsheet ingestion, vector retrieval, cited answers
Vessel segmentation from X-ray angiography for cardiac care, and text-guided facial editing for aesthetic and reconstructive planning.
seen in ✦ coronary artery segmentation, U-Net residual CNN
Analyst agents that reason across category hierarchies, customer profiles, transactions and purchase sessions together, rather than one flat exported table.
seen in ✦ four linked retail datasets, agentic analysis
Workshops for engineering teams, university hackathon mentoring, and written technical material that practitioners actually use.
seen in ✦ Code Wizards 2.0 at SRM, published technical articles
not on this list ? → the constraints usually rhyme. tell me the system and i will say plainly whether i am the right person.
brief → diagnostic → findings → fix → measure
3 questions: what the system does, what it is doing that you cannot explain, and how you would know it was fixed. From that I can tell you whether I am the right person to help.
I read the system, the logs and the data before forming an opinion. Often the reported problem and the actual problem are not the same thing.
The bottlenecks, the cause of each, and what I would change. Sometimes the finding is that the thing you asked for is not worth building, and I will tell you that.
Built by me, or handed to your engineers with enough detail that they can build it. Either way it ships behind a measurement, not behind a demo.
The projection gets checked against production spend and production behaviour, so you can see whether the fix held.
Not to be difficult. Each of these makes the work impossible to judge afterwards, which is bad for both of us.
Every engagement is different enough that a number on a page would be dishonest.