Software engineer · Cloud search · Bengaluru
I build and operate cloud search infrastructure—connecting distributed-systems depth, production empathy, security, and applied research to solve the problems that only appear at real-world scale.
MY SYSTEM MAP / 01
Designing for the edge cases
I’m Priyeash, a software engineer working on OCI OpenSearch. I care about the full life of a system: how it is designed, how it fails, how it recovers, and how clearly we explain it to the people who depend on it.
My work spans management- and data-plane architecture, major-version upgrades, secure identity integrations, live migration, customer-facing debugging, and operational tooling. Alongside engineering, my M.Tech research explores how hybrid lexical-semantic retrieval can adapt its compute budget without quietly giving away relevance.
The connecting thread is simple: go deep, challenge assumptions with evidence, and leave the system—and the team—more capable than before.
Selected themes from my work. Public-safe by design: focused on the problem, the engineering approach, and the customer value.
Live migration
Owning architecture and implementation for remote reindex trust and migration workflows—from API contracts and certificate validation to idempotent rollout, failure recovery, observability, and secure runtime configuration.
Upgrade engineering
Investigated cross-version upgrade paths, fixed a plugin compatibility break, supported time-sensitive mitigations, and improved the guidance teams use to move legacy clusters forward safely.
SAML & enterprise identity
Building domain depth in SAML integrations, reproducing difficult authentication behavior, tracing metadata and role-mapping paths, and turning incident findings into durable product fixes.
On-call systems
Developed tooling for fleet audits, capacity analysis, Kubernetes discovery, health diagnosis, and evidence-backed incident recovery. The goal is not merely faster response—it is safer decisions, clearer handoffs, and less operational drift.
M.Tech dissertation · Information retrieval
My dissertation studies the effectiveness and trade-offs of pre-retrieval gating for hybrid lexical-semantic search in OpenSearch.
The work combines relevance evaluation, systems benchmarking, reproducible experiment design, and adaptive ANN budgeting. Just as important as finding wins: negative results stay visible, test data stays untouched until the protocol says otherwise, and deployment claims stop where the evidence stops.
Local and research results are presented with their scope and guardrails intact.
Trace the real workflow, inspect the actual state, reproduce the edge case. Architecture becomes useful when it explains what the system truly does.
Retries, rollback, observability, and operator clarity are product features. The unhappy path deserves first-class design.
A good result can survive scrutiny. Document assumptions, preserve negative findings, and make the next decision easier for everyone.
Turn one-off fixes into tools, runbooks, tests, and shared patterns so the team compounds what it learns.
Working set
One system. Many moving parts.
I’m always interested in thoughtful conversations about search, distributed systems, cloud reliability, and engineering research.