Currently completing a Master in Management (Financial Markets & FinTech) at NEOMA Business School.
GitGuardian
My role at GitGuardian is to enhance our GTM motion by building the tools and data systems our teams need to be more efficient and intelligent.
-
AI Agent Development: Architected and deployed a suite of AI agents to solve core business challenges. Key projects include an AI Sales Analyst for MEDDPICC deal scoring and a proactive CSM agent that analyzes product adoption signals to detect churn risk and identify upsell opportunities.
-
Internal Tooling: Developed and shipped a suite of internal tools that have a direct impact on team productivity. Key achievement: Built an automation that reduced a critical data processing task for the sales team from several hours to under five minutes.
-
Data-Driven Strategy: Led the data analysis for a major industry event, processing and enriching a +1000-attendee list to create a prioritized targeting strategy that saved the GTM team dozens of hours of manual work.
End-to-End Deep Learning for Portfolio Selection under Systemic Risk Master's Thesis (In Progress)
The Thesis: I am proposing a shift from the traditional "Predict-then-Optimize" paradigm in finance. Instead of forecasting asset prices (which introduces estimation errors), my model uses Deep Learning (TFT and LSTMs) to directly output optimal portfolio weights.
The Goal: The model is trained to maximize the Conditional Sharpe Ratio (CoSR). The resulting autonomous agent acts as a "Doomsday Prepper," optimizing specifically for survival and stability during systemic market crashes rather than just chasing fair-weather returns.
Distinct from my thesis research, this project explores the capabilities of Large Language Models in trading.
An autonomous trading agent framework built to test the viability of LLM-driven decision-making in financial markets. The agent operates in a continuous loop: it ingests real-time market data from the Binance Testnet, enriches it with technical indicators, and constructs a comprehensive prompt for an LLM (deepseek-reasoner).
Governed by a rigid, rule-based constitution, the LLM returns a disciplined trading decision in JSON format, which the system then executes. The entire process is monitored via a real-time Dash control panel.
- Tech Stack: Python, LLM, Prompt Engineering, Dash, Plotly, CCXT, Pandas
- View on GitHub | View presentation site
-
DeepFit - AI Personal Fitness Coach: An AI-powered web app that generates personalized workout plans. A deep dive into systems for analyzing user input and providing data-driven feedback.
-
Alikel Linkgen - Prospecting Automation Tool: A full-stack tool that instantly generates validated, bulk Sales Navigator URLs from a list of company names, dramatically speeding up prospecting workflows.
-
alikel.net - Multi-Agent Chat Application: A custom chat app built to experiment with how specialized AI agents can collaborate within a single interface to assist a user.
- AI & Automation: Gemini API, Deepseek, LangChain, Prompt Engineering, Multi-Agent System Design, n8n.io, Dust.tt
- Full-Stack Development: React, Node.js, Vite, Tailwind CSS, Framer Motion
- Backend & Cloud Architecture: Serverless Functions (Netlify), Supabase (PostgreSQL, Auth, Storage), REST APIs, CI/CD
- Data & Business Ops: Snowflake (SQL), HubSpot, Gong, Apollo
- Web Performance & SEO: Edge Functions, Static Site Generation (SSG), Custom Caching Layers, Core Web Vitals Optimization
- Security: Application Security Concepts, Secrets Detection & Management
My approach is hands-on and problem-focused. I believe the best way to understand a technology's value is to build with it. My goal is to create practical, efficient tools that solve a specific challenge, whether it's streamlining a sales workflow or creating a more effective user experience.