K V Varun Krishnan

1.1K posts

K V Varun Krishnan banner
K V Varun Krishnan

K V Varun Krishnan

@vrn21_

RL @verita_ai | 21 | prev - @hud_evals (YC W25) , https://t.co/dB8ypdwQji, https://t.co/1C2GLdd4LC

Katılım Aralık 2022
1.6K Takip Edilen461 Takipçiler
Sabitlenmiş Tweet
K V Varun Krishnan
K V Varun Krishnan@vrn21_·
I just got laid off — apparently working 7pm to 7am still wasn’t enough. Times haven’t been great lately, but every reset sharpens the stack I've previously built rl envs + infra at Verita and hud (YC W25) stacks: Python, Rust, TS, Docker, AWS DM if you want an engineer who delivers! #buildinpublic #ycombinator #yc
English
46
21
386
55.2K
Krishna
Krishna@almost_krishna·
@vrn21_ Didnt get it, was the layoff related to not working pst hours, or somethin else?
English
0
0
0
341
K V Varun Krishnan
K V Varun Krishnan@vrn21_·
I just got laid off — apparently working 7pm to 7am still wasn’t enough. Times haven’t been great lately, but every reset sharpens the stack I've previously built rl envs + infra at Verita and hud (YC W25) stacks: Python, Rust, TS, Docker, AWS DM if you want an engineer who delivers! #buildinpublic #ycombinator #yc
English
46
21
386
55.2K
Majid Manzarpour
Majid Manzarpour@majidmanzarpour·
@vrn21_ Friendly advice from 25+ years in industry: 1) focus on infra/devops, big gap in market 2) add agents & GCP to your skillset 3) build agentic infra/devops workflows/tools on GitHub Hit me up after you learn
English
4
0
26
1.8K
Rajat
Rajat@rajatsandeepsen·
@vrn21_ its on trending 😭
English
1
0
1
1.1K
K V Varun Krishnan
K V Varun Krishnan@vrn21_·
@PackBropagated It's kind of a gray area, I think expect me to do a 996 pst while being in India ig; this was discussed only during the lay off meeting sadly
English
2
0
6
3.1K
Akash Basudevan
Akash Basudevan@PackBropagated·
@vrn21_ Since when did the amount of time you have given to the job matter?
English
2
0
7
3.5K
Boardy
Boardy@boardyai·
@andrewdsouza @vrn21_ Yeah. I know a founder building RL training envs for web agents, plus an infra lead doing post-training (SFT/RL/evals) stacks. varun, DM me what role you want (systems vs ML infra) + location and I’ll make a couple intros.
English
1
0
2
158
Ekram
Ekram@alelahee·
@vrn21_ CFBR. Small feedback, your resume should not exceed 2 page at all. Best of Luck!
English
2
0
7
4K
Dan
Dan@aidaniil·
member of technical staff or founding engineer for job title?
English
37
0
97
22.2K
Linden Li
Linden Li@lindensli·
We’ve been heads down on our mission to build Specific Intelligence for the enterprise. Today we’re announcing $80M in new funding to help us get there even faster. We’re hiring across engineering and research and working to deploy AI systems that improve the more you use them.
Applied Compute@appliedcompute

x.com/i/article/2041…

English
13
2
89
9.9K
himanshu
himanshu@himanshustwts·
Few pretty interesting trends with Neolab/RL Env companies: 1. The founders are often ex-(frontier lab) or ex-(hedge funds, market-making backgrounds) 2. They stay deliberately lowkey. You are not supposed to see fancy launches. 3. Most of them are being backed by their employers in earlier companies + they are pretty much sorted about who are the potential customers
English
7
1
169
15.3K
seb (sf arc)
seb (sf arc)@hiiinternet·
If you're hiring engineers and want to tap in to eng talent on a weekly basis hmu. These are all the companies on board :) If you're an engineer interested in exploring who wants to chat w/ you check out the link in my bio
seb (sf arc) tweet media
English
9
0
88
4.9K
Saumil
Saumil@OnTheGrapevine·
We’ve raised $4.1M to let AI find the right job for everyone. So today, we’re launching TAL. 3 years and 1M+ users on Grapevine taught us this: The internet today works for companies. Not for talent. Job boards, applications, resumes, ats. The system was built to help companies process people. But people are not pipelines. You are not a resume. You are not an application. You are not a candidate. You are talent. TAL is your talent agent.
English
214
107
1.5K
461.1K
Umesh Khanna 🇨🇦🇺🇸
Umesh Khanna 🇨🇦🇺🇸@forwarddeploy·
Comment below a one liner of the most impressive thing you’ve built (with link) We’re hiring across the board for our teams here at @xai Current students and new grads are highly encouraged!
Baris Akis@barisakis

@tetsuoai @elonmusk @xai 100% we are hiring more and more from people who share their useful projects on X

English
133
12
391
50K
Abhishek Singh
Abhishek Singh@0xlelouch_·
Resume review. I reviewed the resume of an intern today. --- EDUCATION Tier-1 Engineering College | B.Tech in Production Engineering, Minor in Computer Science | Expected 2027 CGPA: 8.08 Coursework: DSA, OS, DBMS SKILLS Languages: C++, Go, Python Backend/Infra: Redis, PostgreSQL, WebSockets, Docker, Kubernetes, Terraform, KEDA Observability/DevOps: Prometheus, Grafana, CI/CD, ArgoCD Cloud/Other: AWS, SQL, Milvus EXPERIENCE AI/ML Intern | May 2025 – Aug 2025 - Built multi-agent LLM workflows using supervisor-style orchestration. - Integrated LangChain ecosystem tools including LangGraph and LangSmith for agent flows and observability. - Improved RAG retrieval quality using semantic chunking and hybrid search. PROJECTS Go-Redis | Dec 2025 - Built a Redis-compatible datastore in Go with a from-scratch RESP protocol implementation, supporting multiple concurrent clients safely. - Implemented 17+ Redis commands including MULTI/EXEC transaction support. - Added persistence with RDB snapshots and AOF log rewriting. - Implemented LRU/LFU eviction and configuration controls for memory and persistence tuning. Go-Jobs | Feb 2026 - Built a distributed job processing system in Go across independent microservices. - Coordinated services using PostgreSQL and Redis Streams with isolated free/premium worker pools. - Achieved effectively-once processing semantics using transactional outbox and idempotent consumers. - Implemented crash recovery using PEL and XAUTOCLAIM for job reclamation. - Deployed on Kubernetes with KEDA autoscaling workers from 1–10 replicas based on Redis Stream lag. - Monitored throughput and scaling behavior using Prometheus and Grafana. Go-Rooms | Jan 2026 - Built a room-based WebSocket server in Go with dynamic join/leave and isolated broadcasts. - Designed a goroutine-per-client architecture with buffered channels and select-based non-blocking message delivery. - Integrated Redis Pub/Sub for cross-instance fan-out and horizontal scaling. - Load tested with k6 at 4,600 concurrent WebSocket connections and 49,574 msgs/sec, with 17ms median connect time and zero failed connections. --- Good: - Strong backend/distributed systems projects. - Good use of Go, Redis, Postgres, Kubernetes, Prometheus, Grafana. - Actual metrics in load testing make the profile stand out. Bad: - Internship bullets are vague. - Skills section was slightly bloated. - Some bullets are too jargon heavy for quick recruiter scans. Ugly: - “Exactly-once” and “proper isolation” are risky claims unless one can defend them deeply in interviews. - “Enterprise-level” says nothing. My Verdict: Very strong student resume for backend roles, but needs tighter wording and a bit less buzzword stacking. Projects carry the profile hard, which is a good thing.
English
16
23
572
61.3K
K V Varun Krishnan
K V Varun Krishnan@vrn21_·
job hunting in 2026 feels totally broken. career pages get ignored, cold outreach is just noise, inboxes flooded. This is my last shot; hoping this reaches the right founder. If you’re hiring or know someone who is, I’d appreciate it if you could point them to my DMs or hello [at] vrn21.com a bit context about me: I'm kv, an upcoming grad, previously interned at hud.ai (YC W25) [built rl envs for code, cua]; mecha.so[ components of a custom os in rust] fileago.com [client side fuse app in rust]. more info abt past over here: vrn21.com/resume built projects like: sandbox for agents(github.com/vrn21/bouvet 100 stars btw); kv store in rust (github.com/hrefto/shorter…); serverless faas plaform with wasm (github.com/orbit-serverle…); native full text search in pg, via an pgrx extension (github.com/vrn21/pdb); distributed command exec (github.com/vrn21/dist-com…); full stack startup success predictor using multi agents(github.com/apricityHQ/apr…). you might find even more projects at my github: github.com/vrn21 tldr; I consider myself somewhat a mix of backend/systems/infra engineer[Rust, Python, Docker, Linux, AWS, Terraform] would love to tackle complex engineering problems at the intersection of these fields.
K V Varun Krishnan tweet media
English
1
0
8
1.4K