Hima Bindu
Yanala
MTS 1, Software Engineer at eBay Inc.
"Beyond the Emergency Backup: Building a Resilient Self-Service Search Architecture with a Dedicated Fallback Index"
"Beyond the Emergency Backup: Building a Resilient Self-Service Search Architecture with a Dedicated Fallback Index"
Wed May 13 - 8:30 AM EDT/New York (See in local time)
Add to Calendar
05/13/2025 8:30 AM
05/13/2025 08:50 AM
America/New_York
#WTGC2025
"Beyond the Emergency Backup: Building a Resilient Self-Service Search Architecture with a Dedicated Fallback Index"
#WTGC2025
"Beyond the Emergency Backup: Building a Resilient Self-Service Search Architecture with a Dedicated Fallback Index"
https://www.womentech.net/ringcentral
https://www.womentech.net/ringcentral
"Beyond the Emergency Backup: Building a Resilient Self-Service Search Architecture with a Dedicated Fallback Index"
"Beyond the Emergency Backup: Building a Resilient Self-Service Search Architecture with a Dedicated Fallback Index"
Don’t miss out and join visionaries, innovators, and thought leaders from all over the world at the Women in Tech Global Conference.
Vote by Sharing
Unite 100 000 Women in Tech to Drive Change with Purpose and Impact.
Do you want to see this session? Help increase the sharing count and the session visibility. Sessions with +10 votes will be available to career ticket holders.
Please note that it might take some time until your share & vote is reflected.
Session: Beyond the Emergency Backup: Building a Resilient Self-Service Search Architecture with a Dedicated Fallback Index
Self-service search is the frontline of customer support, but it's also one of the most fragile. A single traffic spike or partial shard failure can silently return zero results to users, driving costly escalations to contact centers. This talk presents a production-grade architecture for resilient self-service search that treats the fallback index as a first-class reliability mechanism, not an afterthought.Attendees will learn how fan-out amplification can turn a rare outlier request into a widespread failure across large server deployments, and how a layered fallback strategy directly counters this. The session covers a four-component reference architecture spanning content ingestion, a distributed primary retrieval layer, and a dedicated fallback index engineered to significantly reduce CPU time with minimal degradation in retrieval effectiveness.
Routing intelligence is driven by a circuit breaker model with three operational states, closed, open, and half-open, enabling predictable, automated traffic shifting during partial failures. Overload protection through bulkhead-pattern resource partitioning prevents primary saturation from consuming the capacity needed to serve fallback traffic.
On the relevance side, the talk addresses how aggressive automatic synonym expansion can hurt retrieval precision, and why a focused, query-log-validated synonym set delivers the majority of recall benefit at a fraction of the cost. Empty-result avoidance strategies, including tiered title-only and category-level fallback queries, ensure users always receive actionable guidance.
Finally, attendees will see how pairing infrastructure metrics such as tail latency and circuit breaker activation rate with user outcome metrics like query reformulation rate and escalation rate on unified dashboards enables teams to detect relevance regressions before they impact users at scale.
Key Takeaways
- Treat the fallback index as a core part of your search architecture to ensure users always get results—even during failures.
Bio
Hima Bindu Yanala is a seasoned Software Engineer with over a decade of experience designing and delivering enterprise-grade platforms, scalable microservices, and modern REST- and GraphQL-based domain services. Currently a Member of Technical Staff at eBay Inc., she has played a pivotal role in shaping system architecture across shipping optimisation, fulfilment workflows, customer service platforms, and search systems, driving measurable outcomes such as 30–40% API performance improvements and reductions in shipping costs of up to 25%.Her technical expertise spans a broad stack, including Java, JavaScript, TypeScript, React, Spring Boot, Apache Kafka, Elasticsearch, Kubernetes, and AWS, with a strong command of cloud-native and distributed architectures. She is a passionate advocate for automation, having eliminated up to 80% of manual operational effort across the teams she has worked with.
Beyond engineering, Hima is a collaborative leader known for mentoring developers, driving cross-functional initiatives, and contributing to technical direction. Her career spans industries including e-commerce, rail operations, and healthcare, giving her a well-rounded perspective on building reliable software at scale.
She holds an MS in Computer Science from Fairleigh Dickinson University and is a certified AWS Cloud Practitioner and Developer.
