Kathiravan Sengodan

Co-Founder

Alpharetta, Georgia, United States34 yrs 5 mos experience

Highly StableAI Enabled

Key Highlights

Founder of innovative AI/ML infrastructure startup.
Led engineering teams at Salesforce and Oracle.
Holds 11 U.S. patents in distributed systems.

Stackforce AI infers this person is a seasoned leader in AI/ML infrastructure and cloud technologies.

Contact

Skills

Core Skills

Ai/ml InfrastructureSemantic CachingEngineering ManagementCrmPerformance OptimizationCloud InfrastructureMessaging TechnologyMessaging Infrastructure

Other Skills

Vector SearchAI/ML StackOpen SourceDeveloper Community EngagementCosine SimilarityAPI Cost ReductionEngineering LeadershipCustomer Relationship Management (CRM)High Performance TeamsTechnical VisionCost-to-Serve InitiativeSalesforceTest AutomationAgile MethodologiesPerformance and Scale

About

I've spent three decades in software engineering — including years at Salesforce and Oracle building things that scaled to millions of users. At Oracle, where I spent 17 years designing core platform capabilities for WebLogic Server and Oracle Cloud Infrastructure, and at Salesforce, where I led globally distributed engineering teams delivering Marketing Cloud products used by some of the world's largest enterprises. After six years at Salesforce, I took the leap I'd been thinking about for years — and founded Sulci. Sulci is a context-aware semantic caching platform for LLMs that reduces inference costs and latency by intercepting queries, converting them to vector embeddings, and using cosine similarity search to return cached responses. Sub-millisecond cache hits. Meaningful API cost reduction. Open source and developer-first from day one. Building Sulci has been the most rewarding engineering challenge of my career — not because it's the biggest, but because every decision is mine. I hold 11 U.S. patents in distributed systems and messaging infrastructure, with a 12th pending for the core Sulci caching architecture. If you're building high-throughput AI applications and wrestling with LLM inference costs — let's connect. 🌐 https://sulci.io

Experience

34 yrs 5 mos

Total Experience

4 yrs 8 mos

Average Tenure

1 yr 6 mos

Current Experience

Sulci.io

Founder & CEO

Nov 2024 – Present · 1 yr 6 mos

Architected Sulci end-to-end: designed the semantic caching pipeline including query embedding generation, vector similarity scoring (cosine similarity), cache hit/miss orchestration, and cache eviction strategies optimized for LLM workloads.
Built on a modern AI/ML stack leveraging LLMs (OpenAI GPT, Anthropic Claude), embedding models (MiniLM via Sentence Transformers, OpenAI Embeddings), and a pluggable vector backend architecture supporting ChromaDB, Qdrant, Redis + RedisVL, FAISS, SQLite + sqlite-vec, and Milvus Lite.
Drove go-to-market strategy: led developer community engagement, early adopter outreach, and product positioning targeting AI-native engineering teams seeking to reduce LLM API spend and latency in high-throughput applications.
Adopted an open-source-first strategy, releasing Sulci publicly to drive developer adoption, community contribution, and ecosystem trust — accelerating feedback loops and establishing credibility in the competitive AI infrastructure space.
Positioned Sulci as a developer tool first: designed for seamless drop-in integration with minimal configuration, OpenAI-compatible interfaces, and support for multiple LLM providers and vector backends to maximize flexibility for engineering teams.
Focused product strategy on the intersection of cost reduction and latency optimization — targeting high-throughput AI applications where repeated or semantically similar queries make semantic caching a high-ROI infrastructure investment.
Built and published a live demo and comprehensive benchmark suite to showcase Sulci’s cache hit rates, latency improvements, and cost savings across different embedding models, vector backends, and query workloads — enabling developers to evaluate performance against their own use cases.
Managed all aspects of startup operations including cloud cost management, investor-facing technical communications, and IP strategy.

Semantic CachingVector SearchAI/ML StackOpen SourceDeveloper Community EngagementAI/ML Infrastructure

Salesforce

2 roles

Director of Software Engineering

Promoted

Dec 2018 – Oct 2024 · 5 yrs 10 mos

Led the engineering teams, focused on trust, performance, and scale across Marketing cloud product offerings such as Pardot and UMA; that depend on/integrate with Salesforce C360 platform, enabling digital marketers to take full advantage Data Cloud (CDP) for creating effective marketing campaign to generate leads across various channels.
Owned end-to-end people leadership, including hiring, coaching, performance management, and career development of engineers, tech leads, and managers.
Defined and executed the team’s technical vision and roadmap, aligning delivery with organizational strategy, business priorities, and customer outcomes.
Established cross-functional workgroups with product, architecture, and partner teams to drive integration strategy, risk assessment, design reviews, and execution blitzes.
Provided hands-on leadership across DAM/CMS, preference and consent management, CDP, Email/SMS platforms, and subscription management.
Spearheaded a Cost-to-Serve (CTS) initiative, designing a deduplication algorithm that reduced duplicate record inserts by ~70%, improved query performance by ~50%, and delivered approximately $150K in annual AWS cost savings.

Engineering LeadershipCustomer Relationship Management (CRM)High Performance TeamsTechnical VisionCost-to-Serve InitiativeEngineering Management+1

Senior Manager Software Engineering

Dec 2018 – Apr 2022 · 3 yrs 4 mos

Built and led a high-performing engineering team focused on resolving critical performance, scalability, and reliability challenges within Pardot (B2B Marketing Automation), supporting large enterprise customers including Amazon Business, Zoom, and PayPal on AWS.
Defined team charter and engineering best practices with strong focus on Reliability, Availability, Scalability, and Performance (RASP).
Led hands-on efforts to diagnose and remediate large-scale MySQL bottlenecks in high-throughput, multi-tenant environments.
Delivered major scale initiatives including shard switching with foreign-key enforcement, multi-threaded replication, and EC2 canary upgrades, resulting in a 27% increase in replication and write capacity.
Instituted Definition of Ready (DoR) and Definition of Done (DoD) standards and automated functional/performance testing to catch issues pre-release and reduce production incidents.
Founded and led the Performance SWAT (10× Scale) initiative to proactively analyze production metrics and inform capacity planning.
Architected and delivered PerfBot, an automated performance assessment framework that reduced assessment cycle time by 50% and enabled early detection of regressions.

SalesforceTest AutomationAgile MethodologiesPerformance and ScalePeople ManagementEngineering Management+1

Oracle

Principal Member of Technical Staff

Jun 2008 – Dec 2018 · 10 yrs 6 mos

Responsible for design and development of various value added, innovative product features and improvements of Oracle Cloud Infrastructure (OCI & OCI-C) platform.
Designed and built of Nextgen Metering Service that plugs into the Nimbula Control plane (OCI-C)
Service collects and reports the usage of various cloud service (IaaS, PaaS & SaaS) resources to Billing module
Data is collected from NOSQL DB (in the form JSON objects) and the processed data is persisted into ORACLE RAC DB
Service uses Docker based containerization and Jenkins based CI/CD pipeline for build, test and deploy automation
Lead engineer in WebLogic Messaging infrastructure development team, responsible for product design and development from WLS release 6.0 to 12, have performed at various capacities:
Design and implement product features as per the standards and specifications, integrate them with the rest of the product.
Conduct design and code review meetings involving large distributed development team during the product development life cycle (SDLC) to deliver the high quality server product on time.
Write Java and J2EE technology white papers/articles and do presentations on Oracle OTN Web and technology conferences.
As a subject matter expert of JMS Messaging, Persistence, Clustering, HA and Configuration Management, I supported upper stack product teams (such as SOA, BPEL) and Oracle Fusion Application teams with their messaging sub system configuration and deployment as per the configuration best practices that I co-authored with my team member.
Actively inolved in Messaging/JMS related issues from customer support and POCs, to help our Support and Sales organization achieve their goals, indirectly contributing to the product sales and revenue generation.

Python (Programming Language)Java Message Service (JMS)StorageOCICloud InfrastructureMessaging Technology

Bea systems inc

Senior Software Engineer

Nov 2001 – May 2008 · 6 yrs 6 mos

Designed and developed the inter-cluster notification framework, known as CDS (Configuration Directory Service), which is the critical component in WebLogic JMS Distributed Destination Architecture. CDS machinery enables dynamic load balancing of JMS producers, consumers and messages in real time. This is the key enabling technology behind WebLogic MDB and SOA JMS adapters integration.
As a feature owner, designed and developed the Automatic Service Migration for JMS subsystem in WebLogic Server. This feature monitors and detects certain failure conditions (subsystem health) of currently hosting managed Server instance and automatically migrates the JMS subsystem to another available/healthy managed server instance, to achieve high availability and less downtime without the human intervention.
As a lead developer, designed and developed a brand new XML based Modular Configuration framework for JMS sub system in WebLogic Server 9.0 release. This framework helps separate the JMS subsystem configuration into Server specific and Application specific configuration artifacts, and provides flexible ways (administrator configurable system resource, standalone deployable module and packaged with enterprise application archive) to configure the JMS subsystem.

JavaMessaging TechnologyMessaging Infrastructure

Bea systems inc / hcl technologies inc

Senior Software Consultant

Nov 1996 – Nov 2001 · 5 yrs

As a team leader of WebLogic JMS testing effort, responsible for multi-person integrated system applications design, development, execution and automation.

Novell / hcl technologies

Senior Software Consultant

Jan 1995 – Oct 1996 · 1 yr 9 mos

I was a software consultant in Tuxedo quality assurance team, responsible for testing various Tuxedo product APIs by writing client-server applications in a heterogeneous Unix platforms.

Mphasis

Senior Systems Engineer

Jan 1993 – Dec 1994 · 1 yr 11 mos

As a Senior Systems Engineer and team leader worked on Solaris 2.x OS Kernel/Device driver development projects.
As a team member in the Re-engineering/Re-designing project of a Scan Optics Inc., Document Imaging Software, including Format and Language compiler design and development using Motorola M68K assembly language to C/C++ on UNIX System V release 3.2.