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.
As of August 2026: 5 merged, 5 in review, across 4 repositories.
I contribute correctness, crash-safety, and observability fixes to production cache, database, and search infrastructure.
dragonflydb/dragonfly — 5 merged Crash prevention, RSS/OOM test stability, cluster slot-migration correctness, RENAME data-loss prevention.
valkey-io/valkey — in review Fixed replica full-resync failing after restart with AOF-only persistence. Fixed an O(class size × string length) DoS in glob pattern matching.
valkey-io/valkey-search — in review Fixed stale RDB restore counters causing a permanently inflated index metric.
facebook/rocksdb — in review Fixed a native memory leak in RocksJava's thread-local comparator buffer handling.
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.