Profile banner
Siddharth Balyan

Siddharth Balyan

Security researcher and builder β€” ML, observability, and agent systems

sidbing
Bengaluru, India
Joined August 2024

Network

3.8K connections
πŸ”’
Security Engineers
πŸ€–
AI ML Founders
πŸ’»
AI ML Engineers
πŸ›‘οΈ
Cybersecurity Leaders
πŸš€
Startup Ecosystem
πŸ’‘
Product Managers
πŸ›οΈ
IMDEA Institute

Summary

Practical security researcher with hands-on offensive and defensive experience: Siddharth completed multiple security internships (red-team research at FireCompass, security analyst at CloudSEK) and produced tooling and demos (phishing campaigns, Log4j scanner, vulnerability write-ups). His early projects and talks show deep systems and security curiosity. peerlist+1
Builder and OSS contributor focused on agent/LLM tooling: he has contributed to and publicly written about Julep (an agent platform) and Composio, ran evaluations on fine-tuned models, and contributed to open-source repos β€” demonstrating strengths in applying LLMs to practical engineering problems. github+2
Research and applied ML for observability: coauthored a paper (NL2LogQL / "Chatting with Logs") on fine-tuning LLMs to generate LogQL queries and presented this work at Rootconf/HasGeek, showing an ability to combine ML, systems, and security domains. arxiv+1
Public-facing maker and communicator: maintains a personal website (sidb.in), an active X account (sidbing), gives conference talks, and documents projects and postmortems β€” indicating a preference for building in public and engaging with developer communities. Recently published a blog post on designing and experimenting with reinforcement-learning environments (see sidb.in/posts/rl-env-speedrun). sidb+3

Work

Education

Projects

Hobbies

Relearning mathematics for machine learning; writes a Substack and blog posts about learning and systems. sidb+1

Systems and security tinkering: past projects include an IMSI sniffer, container runtime exploration, and offensive security write-ups. sidb+1

Exploring and building reinforcement learning (RL) environments; writing blog posts and sharing experiments about environment design and verifiers. sidb+1