Press any key or click to skip boot checks
A Gen AI & Backend focused Full Stack Engineer.
I like building stuff that actually survives in the wild, from end to end AI infra pipelines, robust backends, data modeling, all the way to making them production ready.
I'm also a low-level programmer, because sometimes you just want to know why the computer is doing that, as that's where real performance comes from.
A little about me ~
AFull-StackDeveloperlivingattheintersectionofMERNandGenerativeAI.Ilikebuildingthingsthatactuallysurviveoutinthewild.AI-poweredappsandinferencepipelines,robustbackends,datatomodeltodeployment,(andeverythinginbetween),Solvingrealproblemsinsteadofdemo-daymagic.Ilikeowningstuffend-to-end,movingfast,breakingandbuldingbackstrongerandbetterthanbefore.Andyeah,I'malow-levelprogrammeratheart,becauseknowingwhythecomputerdoesthatisoddlysatisfying,andit'swhererealperformancecomesfrom.
Cool stuff i've built
A NEURAL FEEDBACK PREDICTION ENGINE POWERED BY META'S TRIBE V2 MODEL TO GENERATE REAL-TIME BRAIN ENGAGEMENT PREDICTIONS AND A 65%+ ACCURATE ATTENTION GRAPH FOR VIDEO CONTENT, ROUTED THROUGH A VLM AND LLM PIPELINE TO DELIVER ACTIONABLE, NON-GENERIC EDIT SUGGESTIONS.
def smooth_neurological_engagement(raw_signals, window_size=5):
smoothed_scores = signal_smoothing_convolution(raw_signals, window_size)
attention_graph = qwen_vision_extract_timestamps(smoothed_scores)
return generate_deepseek_edit_suggestions(attention_graph)A FULL-STACK RAG-POWERED AI PDF WORKSPACE FOR UPLOADING, ANALYZING, SUMMARIZING, AND COLLABORATING ON PDF CONTRACTS. FEATURES TOKEN-BY-TOKEN SSE STREAMING WITH GEMINI 3.6 FLASH, PGVECTOR HNSW SIMILARITY RETRIEVAL, AND MULTI-THREADED REALTIME COMMENTS VIA SUPABASE.
const passages = await matchPdfChunks(queryEmbedding, targetPdfId, 6);
const stream = await ai.models.generateContentStream({
model: "gemini-3.6-flash",
contents: buildRagPrompt(passages, recentTurns, query),
});SMARTCLAUSE PARSES CONTRACTS (PDF/DOCX/TXT) AND EXTRACTS CRITICAL CLAUSES. USING SPECIFIED MISTRAL LLM PIPELINES, IT CLASSIFIES DETECTED CLAUSES BY SEVERITY (LOW, MEDIUM, HIGH) AND PLOTS POTENTIAL VULNERABILITIES.
BLOGS.
Content creation is still a black box, no matter how many analytics you have. You never truly know what your audience feels when they watch your content. And what kind of content is 'popping' at the given time, is still a mystery.
GOT ANY PROJECT OR QUERY ? HIT ME UP!
PROVIDE VALID PARSING STRINGS FOR EXTERNAL COMMUNICATION HOOK.