About
I’m Karthik Balasubramanian, a Software Engineer at Adobe. I’ve worked at Adobe since 2017, starting at Marketo (an Adobe company), with a background in site reliability, tools and infrastructure.
My engineering background spans:
- Production infrastructure & Kubernetes: Running thousands of scheduled, containerized ETL jobs on Kubernetes and Mesos, building GitOps CI/CD with Jenkins and Helm, and maintaining Certified Kubernetes Administrator (CKA) depth.
- Observability & reliability: Logging, metrics, tracing, and alerting across Prometheus, Grafana, OpenTelemetry, and Splunk.
- Agentic systems & evaluation: Shipping LangGraph agents, agent harnesses and loops, researching isolated sandboxes (e.g., Agent Sandbox) for safe code execution, and evaluating agents using the analyze → measure → improve methodology (Hamel Husain & Shreya Shankar).
What I’m working on now
I’m moving down the stack to where LLM cost and latency are decided: GPU kernels and inference engines.
- AI Kernel Engineering with Vizuara AI (in progress): Triton and CUDA kernels, the GPU memory hierarchy, and FlashAttention-style fusion.
- Build a Large Language Model (From Scratch) by Sebastian Raschka, with all my code in pytorch_tutorials.
- Technical writing on kernels, attention, KV cache, continuous batching and speculative decoding, and on serving with vLLM and SGLang. I benchmark each topic the way an SRE would, with p50/p99 latency, time to first token (TTFT), tokens/s and GPU utilization.
I think the gap between kernel engineers and production infrastructure engineers is where inference engineering happens, and I’m building skills on both sides.
Skills
| Area | Tools |
|---|---|
| Languages | Python, Java |
| ML & LLMs | PyTorch, Transformers, LLM evaluation, agentic frameworks (LangGraph), A2A protocol |
| Learning now | Triton, CUDA, vLLM, SGLang |
| Infrastructure | Kubernetes (CKA), Docker, Terraform, Airflow, FastAPI, microservices |
| Observability | Prometheus, Grafana, OpenTelemetry, Splunk |
| Data | PostgreSQL, MySQL, MS SQL Server, Azure Data Explorer, Redis, Databricks |
Education
- Master of Information Systems Management, Heinz College, Carnegie Mellon University (2017)
- B.Tech, Information Technology, SRM University, Chennai (2013)
Certifications
- Certified Kubernetes Administrator (CKA), The Linux Foundation (2024)
- AI Evals for Engineers & PMs, Maven (Hamel Husain & Shreya Shankar) (2026)
- Cloud DevOps Engineer Nanodegree, Udacity (2020)
- Data Engineering Nanodegree, Udacity (2020)
- Machine Learning Engineer Nanodegree, Udacity (2019)
Elsewhere & Contact
I’m focused on ML Inference Engineering and GPU Systems / ML Infrastructure roles where production reliability meets low-level performance. The best way to reach me is on LinkedIn.
- GitHub: karthikBalasubramanian
- LinkedIn: bks4line
- RSS feed