Cortex Crew · DIU
Kawsher HRidoy
Team Lead · ML & Systems
Team lead of Cortex Crew at Daffodil International University (Dhaka). AI and machine learning developer; six competition results in 2026.
Biography
Kawsher HRidoy leads Cortex Crew.
An AI and machine learning developer at Daffodil International University, and the person who leads Cortex Crew from planning through a live demo.
Kawsher HRidoy is an AI and machine learning developer and the team lead of Cortex Crew, a student hackathon and project showcase team at Daffodil International University (DIU) in Dhaka, Bangladesh. He leads Cortex Crew from planning through a live, defensible demo.
In 2026 the team recorded six results under his lead: Champion at the CSAD Project Showcasing Competition, 1st Runners-up at IEEE ICADHI 2026, and Finalist at the AI Innovation Hackathon, the AI Project Competition, the 5th Data Science Summit, and RoboFusion 1.0.
His work focuses on machine learning that ships as production systems. He has led Niro, a patient-owned records platform with Bangla lab and prescription reading; Autopilot, GPU failure prediction and cluster migration on the order of a hundred seconds; and AI Mentor, an eight-signal academic and attendance model with auditable rules. The stack is Python, FastAPI, Next.js, and PostgreSQL, with LightGBM and Azure OpenAI where they fit.
- Python
- FastAPI
- Next.js
- PostgreSQL
- LightGBM
- Azure OpenAI
2026 record
Six results, under his lead.
The team's competition sheet for 2026. Two podium finishes and four finals reached.
Champion
CSAD 2026 — Project Showcasing Competition
Cyber Security Centre, Daffodil International University · Darktrace3
1st Runners-up
IEEE ICADHI 2026 — Project Showcase
IEEE DIU Student Branch · Niro
Finalist
AI Innovation Hackathon 2026 — From Learning to Impact
Dept. of CSE, Daffodil International University · Autopilot
Finalist
AI Project Competition 2026
Dept. of CSE, Daffodil International University · AI Mentor
Finalist
5th Data Science Summit
Daffodil International University · Niro
Finalist
RoboFusion 1.0 — Project Showcasing
UFTB Robotics Club, University of Frontier Technology, Bangladesh · BD Krishibid
Selected work
Systems taken to the stage.
Niro, Autopilot, and AI Mentor are the production systems named in the biography. Darktrace3 is the CSAD Champion build.
Darktrace3
Tells stealer-malware victims exactly what to lock down
Most breach checkers only tell you an email address turned up in a leak. Darktrace3 works from malware and infostealer logs instead, so it can tell a person they were compromised by a stealer at all — and show which sites and accounts were exposed, so they know precisely what to change. Passwords are always masked in results; the tool is built for awareness, not harvesting.
CSAD 2026 · Champion
Niro
Patient-owned medical records, read aloud in Bangla
Patients own their records; an AI reads lab reports and prescriptions in plain Bangla, tracks values over time, and prepares a case summary so a real doctor can review it cheaply. The AI never gives final medical advice — it explains and flags, and consent is a core feature rather than a checkbox.
IEEE ICADHI 2026 · 1st Runners-up
Autopilot
A GPU cluster that heals itself before it fails
Reads GPU telemetry, predicts a failure about 100 seconds ahead, then cordons the node and migrates the running job automatically — a closed loop from prediction to recovery. We measured it against the alert rule real clusters use today, on identical held-out data, and published both numbers.
AI Innovation Hackathon 2026 · Finalist
AI Mentor
Finds the student who is about to fall behind
A deterministic engine scores every mentee across eight academic and attendance signals, then an AI turns the numbers into a plain-language brief. The mentor opens a ranked command centre instead of scrolling a flat table — and every score can be audited back to the rule that produced it.
AI Project Competition 2026 · Finalist