
5
5
France
7+ years of experience
Passionate full-stack developer specializing in React, .NET, secure APIs, and AI integrations (GPT, OpenAI). I develop modern, elegant, high-performance web applications focused on user experience—from frontend to backend. My skills include: • Modern frontend development (React, Angular, TypeScript, Tailwind) • Robust backend development (ASP.NET Core, Entity Framework, JWT, etc.) • AI integration (chatbots, intelligent assistants, GPT APIs, etc.) • SQL databases, dashboards, and secure systems

SignalWatch AI helps enterprise GTM, finance, strategy, and compliance teams unlock live public web data for decision-making. Users enter a target company, its website, competitors, a hackathon track, and a monitoring goal. The .NET 10 backend builds targeted search queries and uses Bright Data SERP API to collect live public web results. The backend filters low-value search/navigation results, classifies useful signals into GTM, Finance / Market, or Security / Compliance categories, and sends the structured signals to Gemini AI. Gemini generates an executive summary, risk score, opportunity score, and recommended business actions. The React dashboard displays the final report with source links, confidence scores, export-to-Markdown, and copy-summary actions. The result is a practical intelligence workflow: live public web data → AI analysis → executive-ready report.
31 May 2026

AgentOps 360 is an autonomous enterprise operations agent that transforms unstructured business input into structured execution plans. In many companies, meetings create a lot of information, but decisions, tasks, risks, owners, and follow-up messages are still handled manually. AgentOps 360 solves this by turning meeting notes, uploaded documents, PDFs, and audio transcripts into a clear business dashboard. The system uses a multi-agent workflow with five specialized agents: an Intake Agent, Reasoning Agent, Planning Agent, Communication Agent, and Audit Agent. Together, they analyze the input, extract decisions, generate action tasks with owners and priorities, detect risks with mitigation strategies, draft professional follow-up emails, and create an auditable agent trace. AgentOps 360 is powered by Gemini for reasoning and structured analysis, Speechmatics for audio transcription, PdfPig for PDF text extraction, React and TypeScript for the frontend, and ASP.NET Core with SQL Server for the backend and persistence. Each analysis is saved in the database, allowing users to reopen previous runs, track execution history, and export a structured executive report. The app also includes a fallback engine to keep the workflow reliable if the AI provider is temporarily unavailable. AgentOps 360 is designed for managers, founders, operations teams, HR teams, sales teams, and customer support teams who need to move faster from conversation to execution.
19 May 2026

BobGuard DevFlow is a repository-aware delivery command center powered by IBM Bob. Modern developers often receive messy bug reports, feature requests, refactoring goals, security concerns, or performance issues, then spend time manually transforming them into implementation plans, tests, risks, and pull request notes. BobGuard DevFlow turns that unclear input into a structured engineering delivery package. The app supports five realistic developer scenarios: Bug, Feature, Refactor, Security, and Performance. Each scenario generates a specific result with analysis metadata, scenario label, confidence score, estimated time saved, executive summary, repository impact map, implementation plan, test strategy, risk radar, pull request pack, IBM Bob workflow, recommended next Bob prompts, and exportable Markdown/JSON reports. IBM Bob was used throughout the project as an AI development partner. Bob helped review the backend architecture, frontend cockpit design, output quality, demo mode, final demo flow, repository structure, and testing validation. The repository includes exported IBM Bob reports in the bob-reports folder as evidence. BobGuard DevFlow is not another chatbot. It is a practical workflow layer that turns IBM Bob-assisted repository reasoning into repeatable engineering artifacts that can be reviewed, shared, exported, and attached to a real development process.
17 May 2026

AgentWatch is a real-time AI agent security platform built on Veea Lobster Trap, securing both Google Gemini and Anthropic Claude backends equally. THE PROBLEM: AI agents read files, hit APIs, and trigger actions in production — but guardrails haven't kept up. A single malicious prompt can leak credentials or exfiltrate data. Most teams have zero visibility into what their agents are doing. WHAT WE BUILT: Deep Prompt Inspection: Every prompt analyzed for injection attempts, jailbreak signatures, PII, credential leakage, and exfiltration patterns via Lobster Trap DPI. Risk scores 0.0–1.0 in real-time. Multi-LLM Security: AgentWatch secures both Gemini 2.0 Flash, Gemini 2.0 Pro, and Claude Sonnet. The same Lobster Trap security layer applies to all backends equally — switch models without losing governance. Policy Enforcement: First-match-wins YAML rules compatible with Lobster Trap. Actions: ALLOW, DENY, LOG, HUMAN_REVIEW, QUARANTINE, RATE_LIMIT. HIPAA, SOC2, and finance packs included. Real-Time Dashboard: Blazor WebAssembly plus SignalR — every attack and alert appears live. Risk trend charts, threat breakdowns, live event feed. Full Audit Trail: Every interaction logged with declared vs detected intent, risk score, and threat flags. CSV export for regulators. LIVE DEMO RESULTS: - Prompt Injection: 0.95 risk → BLOCKED - Data Exfiltration: 0.98 risk → BLOCKED - PII Detected: 0.55 risk → LOGGED - Intent Mismatch: 0.78 risk → HUMAN_REVIEW - Clean Request: 0.02 risk → ALLOWED STACK: ASP.NET Core 10, Blazor WebAssembly, SignalR, SQLite, Google Gemini API, Anthropic Claude, Veea Lobster Trap. Lobster Trap is the floor. AgentWatch is the ceiling.
19 May 2026

ContentDNA is a multimodal AI application that generates a unique creative fingerprint for any media content. Just like DNA reveals a person's identity, ContentDNA reveals the hidden creative identity of your work. Upload an image, video, or audio file and instantly receive a deep structured analysis: a DNA Score (0–100 distinctiveness rating), a 5-dimensional mood profile (energetic, optimistic, cinematic, melancholic, intense), pacing and rhythm analysis (cut rate, BPM, visual density), a style fingerprint with 8 descriptive tags, color palette breakdown, the top 3 real-world creative references it most closely resembles with match scores, target audience prediction (age range, lifestyle, best platform), and a plain-language AI creative brief. The application is built on a React 18 + Vite frontend with a Node.js + Express backend. Media preprocessing uses ffmpeg for video frame extraction and Sharp for image optimization. The AI vision layer uses Gemini 1.5 Flash for multimodal understanding across image, video, and audio inputs — architected to run on AMD Instinct MI300X via ROCm and Qwen-VL when AMD Developer Cloud credits are provisioned. ContentDNA targets filmmakers, content creators, brand teams, and audio producers who need to understand the emotional and stylistic DNA of their creative work in seconds. The project is fully open-source and deployed as a Hugging Face Space.
10 May 2026