eRTMAC-NWIS (Nearby Wells Intelligence System): An AI-Powered Offset Well Knowledge and Decision Support Platform for Drilling Operations
Oil India Limited
Good pick. Genuinely. Now sit down, because the judges are going to try anyway — and this is what they will try.
Strong pick. One of the better-specified industry problems in this block — build on the public Volve or Norwegian well reports, make the depth-triggered alert deterministic, and be candid that the corpus is foreign. Roughly 85–190 teams are expected to go here.
The receipts
Every red flag on this statement, in full. These are the four places it bites.
Exhibit A
No Oil India well data is public, so your corpus will be North Sea wells with different formations and nomenclature, and you should say so rather than implying Indian coverage
It gets worse
Historical drilling reports are inconsistent scanned documents spanning decades of formats, and extraction accuracy on the oldest ones will be poor
Still reading?
Deciding which offset well is genuinely analogous requires geological judgement — proximity alone is misleading when formations pinch out — and a drilling engineer will press exactly there
And the finisher
A false alert that cries wolf at every depth trains the crew to ignore it, so alert precision matters far more than recall here
The damage report
Every score this statement earned, and what each one actually costs you.
Feasibility
3/5Buildable. Not comfortably. There is a week in here you have not planned for yet.
The extraction and mapping stack is standard, and public well report archives exist to work from — Equinor's Volve dataset and the Norwegian and UK public well repositories publish real completion reports — but no Indian field data is available so your corpus will be foreign wells with different formation names.
Innovation scope
4/5There is something genuinely new here. Do not bury it under another dashboard.
How you extract loosely structured operational events from decades of inconsistent report formats, and how you decide that an offset well's experience is relevant to the active well's current position, are both genuinely open problems.
Clarity
5/5The ask is unambiguous, which quietly removes your favourite excuse.
The description numbers the four current gaps and then numbers the required capabilities against them, naming the exact event types to correlate — mud losses, kicks, stuck pipe, casing, cementing — so nothing about the deliverable is left to interpretation.
Acceptance potential
4/5Strong footing before you have written a line. Try not to waste it.
Genuinely underrated — the requirement is precisely enumerated, the demo is immediate, and drilling operations intimidate most student teams so the field will be thin, though you must be upfront that your corpus is foreign well reports rather than Oil India's own.
Effort
HeavyHeavy. Somebody on this team is not sleeping in week three. Pick who, on purpose.
The document extraction pipeline, the structured event store, the geospatial correlation view and the depth-triggered alerting are four connected pieces, with extraction from messy scanned reports absorbing most of the time.
Demo-ability
EasyEasy to demo — and so is everyone else's. Working is the floor here, not the achievement.
A map with offset wells and an alert firing as the bit approaches a known trouble zone tells the whole story in one interaction, and the value is obvious without any drilling knowledge.
Data
None suppliedNo dataset comes with this one, so every accuracy figure you quote is a number about labels you invented.
Nothing is provided with the statement. You are sourcing, cleaning and labelling it yourself, and that work is invisible in the demo but very visible in the questions.
The demo they will have already seen
Somewhere around 85–190 teams are heading here, and the description is doing the choosing for most of them. They will read the same brief, reach the same architecture, and build a version of the same demo you are planning. Being correct is the floor. If your five minutes could be swapped with the team before you and nobody in the room would notice, you have not picked badly — you have built predictably, which costs exactly the same and hurts more.
What survives
The ground worth standing on when the questions start.
- Equinor's Volve dataset and the Norwegian and UK public well archives publish genuine completion and daily drilling reports, so you can build on real documents rather than invented ones
- The alert is deterministic — a spatial query plus a depth comparison — so the headline feature works reliably without depending on model accuracy
- The description names the exact event categories to extract, which gives your extraction schema a defensible origin
Nothing here is fatal. It is just the list of places this statement pushes back, and you now get to push there first.
The framing is a joke. The findings are not — they are the same analysis on the statement page, and every line above is attached to a score or a fact in the record. It is one opinion with its reasoning attached, so argue with it before you trust it.