Software Engineer II @ American Express
Distributed Systems • Databases • Caching • AI Infrastructure
I'm a software engineer focused on the infrastructure behind modern software systems — distributed systems, databases, caching, and AI infrastructure.
I like understanding systems from first principles: how they behave under load, why they fail, and how specific engineering decisions trade off performance against reliability. I explore these questions through hands-on open-source contribution, reproducible experiments, and writing.
I contribute to production infrastructure projects, primarily around correctness, crash-safety, reliability, and performance in databases, caching, and distributed systems.
dragonflydb/dragonfly
- Crash prevention and stability fixes
- Cluster slot-migration correctness
- RENAME data-loss prevention
- RSS/OOM test stability
- Replication and storage-related fixes
BetterDB-inc/monitor
- Reliability and correctness improvements
I also have many open pull requests and ongoing contributions across other infrastructure repositories, including Valkey, Valkey Search, and RocksDB, covering areas such as replication, persistence, pattern matching, index correctness, and native memory management.
→ All pull requests → Dragonfly commits → BetterDB Monitor commits
An experimentation platform for distributed systems.
SlimyBug makes distributed-systems failure modes — retry amplification, cascading failures, admission control, connection-pool saturation, overload behavior — observable through controlled, reproducible experiments, so they can be studied directly instead of learned from production incidents.
Current research areas Database latency propagation • Retry amplification • Circuit breakers • Retry jitter • Connection pool capacity • Admission control • Signal freshness • Overload onset • Admission deferral • Connection pool self-locking
Long-term direction Research-grade, reproducible experiment methodology; an experiment DSL & CLI; evidence-backed engineering findings; a learning platform for distributed systems.
Stack: Python • PostgreSQL • Docker • Prometheus • Grafana • OpenTelemetry
Learned semantic cache eviction for LLM applications.
Investigates lightweight, learned eviction policies against classical heuristics (LRU, FIFO, GDSF) using reproducible experiments on real conversational workloads.
Distributed Systems • Databases • Caching Systems • AI Infrastructure • Performance Engineering • Reliability Engineering • LLM Infrastructure • Developer Infrastructure
I write about the experiments, investigations, and findings behind my systems work — distributed systems, databases and caching, AI infrastructure, performance, reliability, and software engineering more broadly.
📚 Medium
Languages: Python • Go • C++ • JavaScript Backend: FastAPI • Flask • PostgreSQL • Redis • Kafka • RabbitMQ Infrastructure: Docker • AWS • GCP • Prometheus • Grafana • OpenTelemetry AI: OpenAI • Claude • LangChain • Hugging Face • pgvector
Cache design & eviction algorithms • Consensus & replication protocols • Distributed systems internals • AI serving infrastructure • Experimentation frameworks
🌐 Portfolio • 💼 LinkedIn • 📝 Medium • 📧 shikha.py36@gmail.com
Build systems. Run experiments. Produce evidence. Share findings.