AI engineer building production systems with LLMs, RAG, & MCP. Co-founder @hdgelabs. OSS contributor: Zed, OpenAI, CURL. Built GitHub issue solver MCP server.
Just got a PR merged in @curl using an AI-powered, repo-aware analysis tool I built...
it found the exact file & pattern in 30s. DM me if you want a 20-min demo. #opensource#devtools
Spent the weekend building a workspace consistency audit for a 1,100-integration MCP infrastructure repo. Surfaced 251 OAuth integrations
with inline scopes (architectural drift), 1,119 Zod schemas missing AI-discoverable descriptions, and 584 raw error throws. @MetorialAi
One pattern I've been noticing in AI adoption: For years, a large part of learning happened in public. People asked questions, Made mistakes, Got corrected, Sometimes got roasted but the process happened in front of other people because learning was not only about getting answers
Stop writing glue code to connect #OTel to #ClickHouse. π
GlassFlow now ingests OTLP data natively: deduplicating, masking PII, and delivering traces/logs/metrics query-ready.
β No #Kafka
β No custom transforms
β Just clean data
See how: glassflow.dev/blog/clean-enrβ¦#OpenTelemetry
Introducing HDGEBot by HDGELabs: Revolutionizing AI customer support for Indian brands!
Features include:
βΈ "Memori" for seamless conversation memory
βΈ Structured ticketing & escalations
βΈ Comprehensive backend dashboard
offering free access for valuable feedback
Contributing to a 15kβ repo taught me something:
Reading code > writing code.
Before touching anything:
β understand patterns
β check past PRs
β follow style religiously
Most rejected PRs fail here, not in logic.
The maintainer told me my PR diagnosis was wrong.
I thought I found a bug. I submitted a fix.
The expensive mistake:
I assumed the user's bug report diagnosis was 100% correct.
The rule I captured:
NEVER assume a bug report diagnosis is correct.
Recent Project: Built an MCP server with semantic search to capture and analyze OSS maintainer feedback patterns.
Impact:
- Prevents common contribution mistakes
- Speeds up PR preparation by 60%
- Demonstrates meta-learning from code reviews
π System Update: Christmas is LIVE! β¨
From late-night debugging to deploying holiday cheer, the GDG On Campus VIT Mumbai fam wishes you smooth builds, zero merge conflicts, and plenty of joy π»β
Strong coffee, clean code, cozy holidays π π
See you next sprint π
#GDGVITMumbai
π TechSprint is here, VIT!
Dive into Open Innovation with GDGoC VIT Mumbai, turn bold ideas into reality, level up your skills, and compete with the sharpest minds on campus. π‘β‘
β³ Last date: Dec 20
Scan the QR to register!
#TechSprint#GDGOCVITMUMBAI#Hackathon#VITMumbai