Platform Software Engineer · Oracle

Kabir Raj Singh

Building agentic LLM systems and petabyte-scale observability platforms on Kubernetes.

About

Platform Software Engineer at Oracle building petabyte-scale log analytics infrastructure and agentic LLM systems on Kubernetes. Published AI researcher (Springer, IEEE) in emotion-aware text generation, with a track record of shipping safe, production-grade LLM interfaces over live operational data. Serial hackathon winner across RAG, computer vision, and fintech challenges — currently focused on agentic workflows for SOC automation and observability at scale.

Experience

Platform Software Engineer · Oracle India Pvt. Ltd. Aug 2024 – Present

Engineered & operate a multi-region, petabyte-scale log analytics platform using OpenSearch, Kafka, and Scala on OCI Kubernetes — ingests 9 PB/day of logs with sub-5s p99 query latency at peak throughput.
Built custom MCPs from scratch for OpenSearch, Kibana, and the log-processing platform for cluster operations & SOC investigations — became the most-used method of accessing the platform for analysis and development.
Designed agentic LLM workflows for automated security rule generation & anomaly detection — cut manual rule creation effort by 70% and identified previously undetected anomaly classes.
Implemented safe, controlled LLM query interfaces over OpenSearch clusters — enabled automated log investigations & root-cause analysis for significant Tier-1 incidents.
Architected advanced observability & alerting pipelines with Prometheus, Grafana & custom metrics exporters — improved visibility and reduced incident triage time across hundreds of microservices.

Software Developer Intern · Tally Solutions Pvt. Ltd. Oct 2023 – Jan 2024

Developed a reimbursement flow application using lightweight ML models for text and image processing built from scratch — streamlined end-to-end workflows.
Recognized as "Budding Tech Genius" for top technical contributions among the intern cohort.

Software Developer Intern · Samsung Research Institute Bangalore Jun 2023 – Jul 2023

Optimized the Android voice assistant wake-up pipeline and ASR processing — significantly enhanced trigger reliability and user interaction.

Research Publications

ECGText: Maximizing GPT Capabilities for Enhanced Text Generation and Emotional Insight Springer

Designed an emotionally-aware text generation pipeline integrating GPT models with sentiment and emotion classification — evaluated using cosine similarity, winnowing, POS distribution analysis, and sentence structure statistics.

ECGText: Crafting Emotional Texts through GenAI IEEE

Developed an emotion-controlled generation framework combining LLMs with RoBERTa (fine-tuned on the Go-Emotions dataset) and an iterative optimization pipeline — achieved superior emotional fidelity and linguistic coherence.

Full research details

Projects

EthicSecure.AI — Compliance-Aware LLM Gateway Demo

Built a production-ready, policy-aware RAG gateway to block unsafe LLM outputs — leveraged LlamaIndex, Qdrant vector DB & semantic routing for low-latency compliance enforcement.

TicketinSite.AI — AI-Driven Product Review Analytics

Created an end-to-end platform aggregating & analyzing reviews from Twitter, Reddit, and Google Play using LLM-powered spam filtering, translation & clustering — produced actionable issue clusters and sentiment dashboards.

ESS2 Claim Surveillance App

Engineered a full-stack application (Next.js frontend + Flask backend) for automated invoice/expense claim processing with OCR — reduced manual data entry by 50%.

Full project details