1stof 85 teams
Hi! Paris H!ckathon
EvAluate: an agentic assistant for teachers and institutions.
MScT Artificial Intelligence at École Polytechnique
Interdisciplinary expert in AI applications and technology management.
I connect AI engineering with technology management, understanding people's challenges and turning them into useful AI solutions.
PDF download pending · Personal statement in Documents

Paris, France
The domains I work across, connected to the work that puts them in context.
Cognition, AI, institutions and innovation.
Editable sample prompt: “What am I reading or exploring outside work?” Personal note to follow.
Object detection · Multi-object tracking · Image analysis · Multimodal systems
Basketball video analytics / Images that talk
Multi-agent workflows · Tool use · Evaluation · Automation
Nabojo / EvAluate / Notion Autopilot
Retrieval · Semantic search · Grounded generation · Evaluation
Nabojo semantic search and evaluation
Python / SQL · Reproducible pipelines · Statistical modeling
Circana automation / GOALIE research
Stakeholder understanding · Market analysis · Strategic recommendations
180 Degrees Consulting / Yopla market entry
Technical languages: Python, SQL, C++ · Areas of work, not proficiency scores.
1stof 85 teams
EvAluate: an agentic assistant for teachers and institutions.
1stof 30 teams
Digital integration of AI in society.
WinnerTilburg University Challenge
Facilitating the first GP consultation in the Netherlands.
Read the university announcement ↗
Nominatedone of 28 students across the Netherlands
A path from cognition to engineering, and from technical possibilities to the needs of people, businesses and institutions.
I moved to Europe to study Cognitive Science and Artificial Intelligence at Tilburg University. I wanted to understand the connections between human psychology and AI, and what each could teach us about the other.
Outreaching Honours, campus involvement, competitions, associations and consulting brought me closer to businesses and institutions. Along the way, I received recognition for my active contribution to campus life.
At École Polytechnique, my focus became clearer: applied AI for businesses, close client contact, building and consulting. As the first technical / AI hire at EdTech startup Nabojo, I began putting those ideas into practice.
“I started by wanting to understand intelligence. Now, I want to put it to work.”
Experience, leadership & education · newest first

One of 11 representatives advising the management board on a European university programme.

11-member board · 7 partnerships · sponsorship +25%.

Reproducible Python behavioural-data pipeline for the GOALIE study, involving 90+ participants.
Python automation across 100,000+ product / pricing points, replacing at least 15 hours of manual work per cycle.

Yopla carbon-footprint SaaS market entry: 6-person team, 15+ competitor benchmarks across 4 countries. Recommended the Netherlands in a 78-page report and client deck.
Where the work happened, in sequence · hover or tap for details
Pins group places that sit close together. Hover or tap a pin for details. Dotted lines follow the sequence in time.
Different applications. A common thread: making technical ideas useful in context.
From game footage to tracked players, team classification and a mapped view of the court.
Collaborative with Sarah Rouphael · Forked repository
YOLO · DeepSORT · Homography

“What is happening in this image?”
A Streamlit image Q&A app combining BLIP captioning, DETR object detection and LLaMA 3 reasoning via Groq.
Computer vision · Multimodal systems
Your critical tasks, in one brief.
Python daily-brief automation that pulls critical Notion tasks, summarises them and emails a digest every two days via GitHub Actions.
Agents & automation · Python / GitHub Actions
Political transparency for France and India. Machine-verified, verbatim quotes from primary sources, not summaries.
Primary-source verification · AI for institutionsConnects homeowners, contractors and material buyers on one platform, combining stackable French public aid with a marketplace for demolition material.
Circular renovation · Community workflows
Official website of MoX, the student body for all MScT students at École Polytechnique. Handles member information, event registration, partnership inquiries and payments.
Community website · TypeScriptHum a melody and the system finds the song from a database of 20 tracks, using pitch extraction and a two-stage DTW match. The project also covers harmonising a hummed tune and generating a music video from a song.
Audio signal processing · PythonPredicting whether a Twitter user is an Influencer or an Observer from a single tweet and its metadata. CamemBERT embeddings, engineered features and a boosted-tree ensemble reached 84.4% accuracy on the Kaggle test set.
Deep learning · NLP · KaggleBehavioural cloning followed by REINFORCE fine-tuning on LunarLander, compared against RL from scratch. Ablations cover demonstration set size, expert quality and covariate shift.
Reinforcement learning · PyTorchAn MVP that flags racism and sexism in text with a BERT classifier, then explains each flag with a Mistral LLM. Backend and frontend run from a single Docker container.
NLP · BERT · MistralClassifies disaster-related social media posts to help triage them, combining the Kaggle Disaster Tweets and HumAID datasets.
NLP · PythonOperate the mouse, change the volume or switch wallpapers using only hand signs and motions captured by a camera.
Computer vision · Human-computer interactionMy solutions to the six computer vision projects in WorldQuant University’s Applied AI Lab: classification, detection, segmentation, facial recognition, medical image generation and text-to-image generation.
Computer vision · PyTorch · OpenCVRead the sample previews below. Downloadable PDFs are not yet supplied.
Prakhar Tiwari
AI applications & technology management
Paris, France · prakhar.tiwari@polytechnique.edu
Education
MScT, École Polytechnique · 2025–2027
BSc Cognitive Science & AI, Tilburg · 2022–2025
Selected experience
Nabojo · AI R&D Engineer Intern / first technical hire
Circana · Technology Consultant Intern
180 Degrees Consulting · Student Consultant
Python · SQL · C++
Computer vision · Agentic AI · Data science
I began with a question about intelligence: how do people learn, and what can AI learn from them? Studying cognition at Tilburg gave me a way to investigate that question.
Consulting and campus involvement widened my perspective. At École Polytechnique, I found a clearer direction: building applied AI in close contact with the people and organisations it serves.
I’m interested in internship opportunities where engineering, thoughtful evaluation and business understanding meet. I want to understand the challenge before deciding what to build.
Sample preview: degree and course overview. Official transcript not supplied.
Sample preview: pipeline notes and collaborative contribution summary.
Sample slot for approved evidence. No certification is claimed.
A second way into the portfolio. Explore the experience, projects and ideas behind the work.
Sample conversation, not a live assistant. Answers draw on portfolio material; unknowns go to direct contact.
How does Prakhar connect technical AI work with business needs?
His experience spans understanding the problem and building the solution. At Circana, Python automation replaced at least 15 hours of manual work per cycle. At Nabojo, agentic checks supported pedagogical review. At 180 Degrees Consulting, he worked in a six-person team on Yopla’s market-entry strategy.
Sources: Experience → Circana, Nabojo, 180 Degrees Consulting
Hiring, building or exploring a useful application? I’m interested in the conversation.
For a quick overview: CV sample preview · Experience · Selected work