Technology & funding partner

Evolution Cloud Services is funding Escape Velocity — the same enterprise AI and cloud engineering problems their teams solve every day.
Hackathon 1.0 · EVOCS-funded · Velloe AI Innovation Center, Noida
Hackathon 1.0
The speed it takes to break free and launch. One day to build with document intelligence, enterprise RAG, and multi-agent systems — the same problems EVOCS and Velloe engineers work on. Shortlisted builders get interviewed for a Velloe internship.
01
Register
Resume + LinkedIn
02
Build solo
9 AM – 6 PM on-site
03
Pitch video
Show your work
04
Internship
Top builders interviewed
Funded by EVOCS (Evolution Cloud Services) · Hosted by Velloe
Date
Sat, Sep 26, 2026
Time
9:00 AM – 6:00 PM
Venue
Velloe AI Innovation Center, Noida
Who's building
100 individual builders
Why this exists
EVOCS (Evolution Cloud Services) is funding Escape Velocity, run out of Velloe's AI Innovation Center in Noida — the hub behind AI-ready infrastructure and enterprise AI systems. This is the first time that floor opens to builders from outside the company for a day, not to judge a demo, but to find people who think the way our engineering team does. The best individual performances get carried straight into Velloe's internship pipeline: interview, then offer.
Sponsors & partners
Technology & funding partner

Evolution Cloud Services is funding Escape Velocity — the same enterprise AI and cloud engineering problems their teams solve every day.
Host · AI Innovation Center

Velloe opens the AI Innovation Center in Noida for the first external build day — mentorship, infrastructure, and the path to a Velloe internship.
Campus outreach network
Escape Velocity is open to final-year and postgraduate builders from Delhi NCR campuses — ready to build solo for a full day.
B.Tech
4th-year students
MCA
MCA candidates
M.Tech
M.Tech candidates
PhD
Candidates in AI, ML, data science, or cybersecurity
Problem tracks
Hover a track for how to think, solve, and what to ship in your pitch archive.
“One goal, many minds.”
Build a multi-agent system (planner, retriever, executor, validator) that takes a complex enterprise request from intake to resolution with no human in the loop until final sign-off.
Track 01 · Agentic architecture
Hover for how to think & solve →
How to think & solve
Decompose the request into roles before you write code. Define handoffs, failure modes, and where a human must sign off. Prove the loop closes on one realistic enterprise scenario end to end.
What to ship & archive
Working demo, agent-role diagram, one full trace from intake to resolution, and a pitch on what you would harden next.
Track 01 · Agentic architecture
“Every answer, sourced.”
Build a retrieval-augmented assistant over messy enterprise documents that answers natural-language questions with citations back to the exact source.
Track 02 · Enterprise RAG
Hover for how to think & solve →
How to think & solve
Start with messy documents, not clean samples. Prioritize citation fidelity over fluency — if you cannot point to the source, do not answer. Test with questions designed to break naive retrieval.
What to ship & archive
Query UI with inline citations, eval notes on 5+ hard questions, and a short write-up of chunking and embedding choices.
Track 02 · Enterprise RAG
“Relationships are the answer.”
Model entities and relationships as a knowledge graph, combining graph traversal with vector retrieval to answer multi-hop questions flat RAG misses.
Track 03 · Graph RAG
Hover for how to think & solve →
How to think & solve
Map entities and edges first; retrieval is graph walk plus vectors, not either/or. Design one multi-hop question flat RAG cannot answer — then make yours answer it with evidence.
What to ship & archive
Mini knowledge graph, visible hybrid retrieval path, and one live multi-hop win demonstrated in the pitch.
Track 03 · Graph RAG
“No legacy. No excuses.”
Design and prototype a new healthcare product from a blank slate (patient intake/triage copilot or clinical documentation assistant), with privacy and auditability built in from day one.
Track 04 · Healthcare greenfield
Hover for how to think & solve →
How to think & solve
Treat privacy and auditability as product features, not footnotes. Pick one workflow, scope ruthlessly, and document every data touchpoint before you generate content.
What to ship & archive
Clickable prototype, one-page privacy/threat sketch, and an explicit audit trail for a single user journey.
Track 04 · Healthcare greenfield
“From scanned to searchable.”
Ingest unstructured or scanned documents (forms, invoices, clinical notes) and extract structured, verifiable data from handwriting, tables, and inconsistent layouts.
Track 05 · Document intelligence
Hover for how to think & solve →
How to think & solve
Assume scans are ugly — skew, noise, handwriting. Extract structure you can verify field by field; flag low-confidence extractions instead of silently guessing.
What to ship & archive
Before/after on 2–3 document types, structured output you can validate, and the failure cases you caught.
Track 05 · Document intelligence
“Catch it before it spreads.”
Detect anomalous behavior across logs, network traffic, or user activity, and explain the reasoning to an analyst.
Track 06 · AI in cybersecurity
Hover for how to think & solve →
How to think & solve
An anomaly without explanation is noise. Tie every alert to evidence in logs or traffic; optimize for analyst trust, not alert volume.
What to ship & archive
Detection on sample telemetry, a reasoning panel per alert, and one false positive you deliberately avoided.
Track 06 · AI in cybersecurity
How the day runs
Problem statement drops at check-in
You won't see it in advance — everyone starts from the same line.
9 AM – 6 PM build window
One day, on-site, at the AI Innovation Center.
Record a short pitch video
Walk the jury through what you built and why.
Jury scores every submission
Velloe engineers and invited judges evaluate each video.
Mentorship
10 pods × 10 builders
No teams to assemble — every builder works solo, but you're never on your own. 100 builders split into 10 mentor pods, each guided through the day by a dedicated Velloe mentor.
The path to an offer
01
Build
9 AM – 6 PM. One problem statement, solo.
02
Pitch video
Every builder records and submits a short walkthrough.
03
Jury review
Judges score each pitch against the day's rubric.
04
Pod shortlist
Top 3 from each of the 10 pods move forward — 30 in total.
05
Internship
Shortlisted builders interview, then the best get an offer.
What's covered
Everyone
Goodies
Every registered builder walks away with Escape Velocity swag.
Meals
Lunch & snacks
Full meals and snacks through the day, on Velloe.
Winners
Prizes
Gifts for the top performers, judged on the day's rubric.
Mentors & jury
Mementos
A keepsake for every mentor and judge who gives their day to Escape Velocity.
Escape Velocity Registration
Registration closed
Registration for Escape Velocity Hackathon 1.0 is now closed. Thank you for your interest — we look forward to seeing builders on event day.