AI & ML engineering · LLM systems · Applied AI

Engineering AI systems from data to inference.

I’m Yeshwanth Janne. I build and explore language-model systems, AI-powered applications, and the backend platforms behind them — with a focus on practical implementation, measurable behaviour, and reliable software foundations.

Explore AI projects Connect on LinkedIn ↗

Primary interests LLM engineering · Model serving · Applied AI · AI platforms

LLM systemsTokenization, decoder-model foundations, local inference

AI applicationsUseful user workflows with transparent AI assistance

AI platform foundationsDocument pipelines, data governance, APIs and access control

AI-first portfolio

LLM systems, AI products
and platform engineering.

The work below is ordered by relevance to AI and machine-learning engineering roles: model systems first, followed by applied AI products. Each project is labelled by its actual stage.

MIZANEnterprise AI platform

AI platform · Data governanceOngoing engineering

MIZAN — AI lifecycle & governance

An enterprise platform designed to bring structure to AI data workflows, experimentation, evaluation, governance and runtime operations. My hands-on engineering has focused on the platform foundations that make those workflows secure and manageable.

Implemented engineering areas: multi-tenant identity and access foundations, secure document ingestion, storage abstraction, bounded file parsing, dataset models, API endpoints and supporting frontend workflows. The broader product architecture covers governed datasets, evaluation, artifact lineage, retrieval and runtime operations.

PythonDocument ingestionDataset workflowsRBACAPI design

Explore MIZAN ↗

MYLLM / AetherLLM systems research

Language models · InferenceResearch prototype

MYLLM — building an LLM stack

A CPU-first language-model engineering project exploring the components around a small decoder model: byte-level BPE tokenization, transformer primitives, dataset-to-checkpoint training workflows and local inference.

Engineering work: OpenAI-compatible local API serving, tokenization utilities, sampling and search strategies, inference telemetry, batch scheduling and checkpoint validation. Aether is a 3.22M-parameter untrained model core — a systems prototype, not a production-trained foundation model.

PythonPyTorch CPUByte-level BPETransformer decoderInference APIs

View GitHub profile ↗

AI ResumeApplied AI product

Applied AI · Information extractionLive product

Free AI Resume Builder

A live AI-assisted career tool that turns a candidate’s existing resume and a target job description into clearer, more relevant application material while keeping suggestions grounded in the candidate’s stated experience.

Product workflow: resume upload and extraction, job-description analysis, keyword and skills-gap guidance, factual rewriting, resume preview, and PDF/DOCX export. Its ATS-readiness score is an estimate, not a prediction of any employer’s proprietary ATS result.

AI-assisted writingDocument extractionJob-description analysisPDF / DOCX

Try the builder ↗

ilamok AIAI assistant application

Applied AI · Cross-platformProject site live

Ilamok AI

A cross-platform AI assistant product for Android and Windows. The product direction is to make AI interactions accessible through a focused interface, with an emphasis on practical everyday use rather than a research-only experience.

Focus: AI product experience, application delivery and simplifying access to assistant capabilities. Platform availability and implementation details are described on the project site.

AI applicationAndroidWindowsProduct engineering

View Ilamok AI ↗

How to read this portfolio: live product availability does not automatically mean every underlying AI capability is production-grade. MIZAN is ongoing platform engineering; MYLLM/Aether is a research prototype whose model has not been trained into a production foundation model. I distinguish implemented work from broader product direction so the technical discussion stays precise.

Additional product engineering

Web platforms & client projects

These projects show the broader product and software delivery experience behind my AI work — building usable interfaces, workflows and backend-connected digital experiences.

hostayleLiving platform

Marketplace · OperationsLive product

Hostayle

A PG, hostel and co-living platform connecting residents with available stays, alongside owner workflows for properties, rooms, residents and rent operations.

Web platformProperty workflowsResident experience

Visit Hostayle ↗

KakatiyaHealthcare

Healthcare · Web experienceLive website

Kakatiya Hospitals

A healthcare web experience for discovering specialties, doctors and hospital services, with appointment and emergency contact journeys.

Healthcare websiteInformation architecturePatient journeys

Visit Kakatiya Hospitals ↗

EDISIONBusiness website

Commerce · Product catalogueLive website

Edision India

A product-led website presenting electrical wires, cables and fuses, company information and customer enquiry paths.

Product discoveryCatalogueBusiness web

Visit Edision India ↗

AromaHospitality

Food · HospitalityLive website

Aroma Indian Grill

A restaurant website bringing together menu discovery, online ordering, reservations and catering enquiries.

Restaurant websiteOnline orderingCustomer experience

Visit Aroma Indian Grill ↗

Some client-facing projects are associated with HENNT Technologies and may involve collaborative delivery. I can explain my specific role, implementation choices and project status during a technical discussion. Additional published work is available in the HENNT portfolio ↗.

AI engineering approach

Beyond a model call.
Build the system around it.

I’m interested in the engineering that makes AI useful in real products: data quality, model behaviour, inference constraints, secure workflows and clear user outcomes.

01 / DATA

Prepare data carefully

Work with document ingestion, parsing, structured datasets and privacy-aware foundations before data reaches downstream workflows.

02 / MODELS

Understand the model path

Explore tokenization, decoder-model components, training and checkpoint paths, sampling, and local model-serving interfaces.

03 / EVALUATION

Measure, then improve

Prefer explicit validation, repeatable tests and observable inference behaviour over unsupported claims about model quality.

04 / PRODUCT

Keep humans in control

Design AI features around clear user intent, truthful source information, reviewable output and understandable product workflows.

Software engineering foundation

Enterprise .NET backend engineering

Professional work on enterprise account-management APIs, including update flows, command and handler patterns, request validation, business-rule alignment and unit tests. This experience strengthens the backend discipline I bring to AI platforms and applications. Internal business details are kept confidential.

C# / .NETREST APIsFluentValidationUnit testingCode review

What I am pursuing

AI engineering roles grounded in software fundamentals

I’m particularly interested in LLM engineering, model-serving and inference systems, applied AI products, and AI platform/backend engineering — roles where experimentation is paired with careful implementation and verification.

Hands-on buildingLearn by implementing components and connecting them into working systems.

Technical honestySeparate prototypes, implemented capabilities and production-ready systems.

Engineering qualityUse clear contracts, validation, tests and maintainable boundaries.

Continuous learningImprove through benchmarks, code review, documentation and practical feedback.

Technical toolkit

AI systems first.
Strong foundations underneath.

A practical toolkit shaped by the projects above. I focus on the technologies I have worked with directly and continue to deepen my knowledge through implementation.

LLM & model systems

PythonPyTorch (CPU)Byte-level BPETransformer decoder fundamentalsTraining & checkpoint workflowsSampling & searchLocal inferenceInference benchmarking

Applied AI & data workflows

AI product developmentDocument ingestionText extraction & parsingDataset workflowsJob-description analysisAI-assisted writingFastAPIOpenAI-compatible APIs

Backend & platform engineering

C# / .NETREST API designSQL / MySQLMulti-tenant systemsAuthentication & RBACRequest validationUnit testingGit & code review

Let’s connect

Let’s build AI that solves meaningful problems.

I’m pursuing AI/ML engineering opportunities across LLM systems, model serving and inference, applied AI products, and AI platform engineering. I welcome conversations about research prototypes, production systems and challenging engineering problems.

Connect on LinkedIn ↗ Explore GitHub ↗