TEJAS KUMAR

Led backend optimization initiatives at HackerRank, enhancing platform performance and developer experience for millions of users.

• Optimized execution runtimes and refined language server protocols, reducing backend latency by 20% and improving code execution speed for millions of candidates.
• Resolved critical container runtime issues using Prometheus + Grafana monitoring on GCP, improving system reliability and reducing incident response times.
• Engineered custom Language Server Protocols (LSPs) for JavaScript and TypeScript, enabling seamless IDE support and reducing candidate environment setup issues by 30%.
• Enhanced PyPy3 runtime by 30% through a custom pycompile pipeline that decoupled compile and execute logic, improving test case throughput.

Contributed to KubeEdge as an LFX Mentee, developing features for large-scale edge cluster management and performance optimization.

• Implemented workload deployment overrides using label selectors in EdgeApplication CRD, enabling fine-grained scheduling for large-scale edge clusters.
• Extended edge controller with `targetNodeLabels`, improving workload placement accuracy across heterogeneous edge nodes.
• Designed a closed-loop flow control system using topology-aware routing, reducing cross-zone latency and improving throughput by 15%.
• Integrated EndpointSlice hints for zone-based load distribution, enhancing traffic balancing in multi-zone environments.
• Upstreamed a core scheduling feature (PR #5845), adopted by the KubeEdge community, demonstrating significant open-source contribution.

Developed critical CI/CD tools and testing infrastructure at Keploy, significantly enhancing developer efficiency and test reliability.

• Developed a Go-based deduplication engine that cut CI test duplication by 70%, saving over 20 developer hours weekly.
• Implemented automatic re-recording for flaky tests, improving test accuracy by 60% and strengthening CI reliability.
• Reverse-engineered MSSQL protocol using Wireshark and built a custom parser, enabling backend teams to simulate DB calls without live databases.

Developed a ServiceAccount Controller to automate security configurations and manage JWT lifecycles in Kubernetes clusters, enhancing security and consistency.

Developed and contributed a feature to KubeEdge, implementing label-based node affinity scheduling for enhanced workload placement in edge environments.

Reverse-engineered the MSSQL protocol and built a custom parser to simulate database responses for backend CI pipelines.