Console online
Dhruv R Krishnan_
Building AI that thinks deeper, reasons better, and ships smarter.
3y 8m
Time in field
6
Roles held
5
Shipped builds
18
Tracked skills
Experience
AI R&D Research Engineer
ActiveDoctreen · Montpellier, France
- Engineer indigenous data pipelines that generate clinically accurate medical data for domain-specific LLM training.
- Design and train a custom LLM using SFT and RL, pushing medical-reasoning performance ahead of ChatGPT on the product benchmark.
- Architect an end-to-end system that builds medically grounded decision trees, helping radiologists produce structured, defensible reports.
- Advance model performance by integrating structured medical knowledge, enabling more accurate, context-aware report generation.
14mo
Duration
32%
Career span
3
Focus areas
Toolset
Model & Research
MDL- LLM EngineeringCore
- Reinforcement LearningCore
- SFTCore
- Neural NetworksAdvanced
- Computer VisionAdvanced
- CNNsAdvanced
- LVMWorking
Systems & Data
SYS- PythonCore
- FastAPIAdvanced
- DjangoWorking
- DatabricksWorking
- MongoDBAdvanced
- PostgreSQLAdvanced
- EDAAdvanced
Web
WEB- Next.jsAdvanced
- JavaScriptAdvanced
- HTMLAdvanced
- CSSAdvanced
Projects
MedReportGen AI
08/2024 — 12/2024AI Engine & Architecture
Pairs Microsoft's Radino vision-language model with an LLM, turning raw clinical data into structured, precise reports.
SWC File Generation from MRI
08/2023 — 02/2024MRI Refining & Graph Generation
Analyzes human brain vascular flow from MRI scans and reconstructs it into SWC graph files.
Code2Graph
11/2023 — 12/2023AI Engine & Front-End
Generates knowledge graphs from source code and auto-derives algorithms — an efficient debugging companion.
Visionary Controls
08/2023 — 12/2023AI Engine & Deployment
A gesture-based control system for computers with voice-activated mode switching.
Lunar Lander via DQN
04/2024 — 06/2024DQN Network & Agent Building
A deep Q-network agent trained to autonomously pilot a lunar lander probe to a safe touchdown.
Recognition
- 01
Winner — MLH Global Hack Week Challenge
Twilio Quest Challenge, international level.
- 02
Runner-Up — CodeSTC National Web Development Contest
Youngest participant at age 15 among graduate competitors.
- 03
2nd Runner-Up — Scinova Science Fest
State-level competitive coding contest.
- 04
Honorary Mention — Spark Hub International ML Hackathon
Project: Heart Failure Prediction Software.
B.Tech, Artificial Intelligence Engineering
Amrita Vishwa Vidyapeetham · 2021 — 2025
8.72
CGPA /10
94.6%
CBSE XII
95%
CBSE X
About me
Building domain-specific LLMs for medical report generation at Doctreen — training with SFT and RL so the models reason more like a clinician and less like a generic chatbot.