Bridging Systems Engineering with Research: Tejas Dhanorkar Driving Innovation In AI-Augmented Automation

At the intersection of enterprise-scale engineering and research-led problem solving, Tejas Dhanorkar is advancing the future of intelligent systems design. With over 12 years of experience in Java full stack engineering and cloud-native application development, Tejas has played key roles in architecting payment platforms, performance-tuned CI/CD pipelines, and building scalable automation frameworks that support compliance, monitoring, and advanced analytics. His industry work is grounded in practical systems thinking, but what distinguishes Tejas is how that technical rigor informs his research contributions-each of which speaks directly to the evolving needs of modern digital infrastructure.

Tejas's core competencies span a wide array of technologies. His expertise in Spring Boot, Kubernetes, containerized deployments, messaging platforms such as RabbitMQ and Kafka, and orchestration tools like Jenkins and GitLab CI/CD has made him a go-to systems engineer for high-impact enterprise solutions. His hands-on involvement in improving build pipelines, optimizing performance for financial transaction flows, and leading engineering teams in geographically distributed environments laid the groundwork for his deeper intellectual engagement with machine learning, semantic AI, and graph reasoning. These competencies directly inform his contributions across three significant research publications that demonstrate how technical depth and domain fluency can intersect to solve pressing problems in automation, security, and law.

Building Developer-First Tooling with Semantic Learning

One of Tejas's most pragmatic research contributions is showcased in the paper "Pull-Request Whisperer: LLM-Enhanced RCA Advisory Bot for Developers," published in Artificial Intelligence, Machine Learning, and Autonomous Systems (Vol 2, 2018). In this work, Tejas addressed a pain point familiar to enterprise engineering teams: how to reduce the rate of regressions entering production through better tooling and feedback mechanisms during code reviews.

The Pull-Request Whisperer introduces an intelligent advisory engine that integrates directly into CI pipelines and version control systems like GitHub. It performs real-time semantic analysis on pull requests using historical failure data encoded as embeddings. This allows it to detect patterns that resemble prior regressions and proactively surface advisory suggestions.
Tejas's contributions to the system architecture were critical. He designed the advisory engine's feedback pipeline, ensuring that alerts are injected into pull requests only when meaningful, and structured in a way that developers find actionable. As stated in the paper, "Tejas says the advisory engine injects feedback directly into pull requests, offering preemptive remediation strategies, testing suggestions, and rollback advice based on real incident data."

This approach reflects Tejas's hands-on experience in leading daily triage sessions, managing incident resolution workflows, and analyzing defect recurrence. His practical understanding of how developers absorb information during code reviews influenced the system's user-facing design. By combining AI-powered insights with human-centric delivery, Tejas was able to create a tool that not only prevents bugs but enhances engineering team efficiency and confidence in CI/CD quality gates.

Detecting Zero-Day Threats through Graph Neural Networks

Another significant research milestone for Tejas is his work on hypervisor-layer threat detection using graph neural networks (GNNs). In the paper "Graph-Neural Threat Detection at the Hypervisor Layer," published in the Edinburgh Journal of Natural Language Processing and AI (Vol 3, 2019), Tejas applied his knowledge of systems architecture, virtualization, and performance engineering to tackle the problem of detecting lateral movement and zero-day exploits in virtualized environments.

This research introduces a lightweight, hypervisor-embedded GNN framework that models inter-VM communication as dynamic temporal graphs. These graphs are used to detect abnormal sequences of communication that may signal the presence of stealthy attacks.

Tejas was responsible for the GNN model design and its performance optimization for real-time environments like VMware ESXi and Microsoft Hyper-V. His prior experience managing transaction flows in high-throughput systems allowed him to make critical decisions around inference latency and CPU load balancing.

As noted in the publication, "Tejas engineered the temporal graph inference pipeline to capture suspicious propagation patterns in real time, enabling the system to detect zero-day exploit footprints with sub-second latency."
Tejas also designed modular integrations to ensure that detection insights could be converted into forensic reports for security analysts-allowing not just detection, but traceable response workflows

Accelerating Legal Argument Drafting through Semantic Retrieval

Tejas's third notable research contribution, "Semantic Precedent Retriever for Rapid Litigation Strategy Drafting," published in the Journal of Artificial Intelligence & Machine Learning Studies (Vol 4, 2020), explores the intersection of legal informatics and natural language processing. This research focuses on enabling legal professionals to rapidly build structured litigation arguments by retrieving and assembling precedent-aligned case information.

The system is built upon dense vector representations using LegalBERT-style embeddings and applies transformer-based generative models to synthesize coherent legal drafts. Tejas's involvement centered around building the backend retrieval system, ensuring that performance requirements for query-to-response time remained under three seconds.
As described in the paper, "Tejas says the generation engine was engineered to align retrieved citations with jurisdiction-specific procedural norms, enabling the system to draft high-quality legal arguments enriched with predictive success insights."

His knowledge of RESTful orchestration, transformer optimization, and latency-sensitive data indexing enabled this system to work effectively in live legal advisory settings. Drawing from his experience with payment compliance automation and document ingestion systems, Tejas introduced architectural safeguards to maintain legal citation integrity and reduce model hallucination risks.

From Engineering Rigor to Scholarly Contribution

Across these research initiatives, Tejas demonstrates a rare blend of engineering rigor and scholarly insight. His research consistently draws upon challenges he has encountered in production environments-whether it's regression detection, security analytics, or compliance automation-and transforms them into structured models that are scalable, explainable, and applicable across industries.

His leadership in major engineering efforts at Discover-such as designing a cloud orchestrator that reduces processing latency by parallelizing external API calls, or creating mock services to decouple test environments from flaky dependencies-reflects the same mindset visible in his research: identify bottlenecks, model them rigorously, and deliver systems that improve reliability and agility.

About Tejas Dhanorkar

Tejas Dhanorkar is a Principal Application Engineer and AI-enabled systems researcher with over a decade of experience designing enterprise applications, intelligent CI/CD pipelines, and real-time monitoring systems. His work in Java full stack development is complemented by his specialization in Spring Boot, Kubernetes, and AWS/PCF cloud ecosystems. At Discover Financial Services, he has led foundational efforts in payment network optimization, RCA automation, and orchestrated parallel test execution-one of which earned him the Discover President's Award. Tejas's research spans semantic legal AI, developer intelligence tooling, and graph-based threat detection, demonstrating his ability to bridge operational engineering with scholarly innovation.