Vahid Faraji
🤗
Researcher-Builder

Applied AI
from research to production

I build the agent systems enterprises actually put into production.

Open to roles Open to AI Solutions Architect, Data Product Engineer, AI Platform Engineer, Agentic AI Engineer and R&D Researcher roles. Book a call
the 30-second version
Multi-Agent Systems Context Engineering MCP Protocol Applied AI Open Source AI SDK Kits Published Research
Published at ACM CAIS '26 / aX Tokalator / aX Skill Extraction
Download CV See work Portfolio Work with me Suggest a time
9+
Years shipping
95%
Agent accuracy
72.5%
Faster classification
60%
Workload reduction
3+1
Publications + Patent
400+
People trained
Istanbul --:--
You --:--
awesome-applied-ai · July 2026

A working map of applied AI for the enterprise

Six layers: Guardrails and Governance, Evaluation and Observability, Orchestration, Caching, Memory, and Retrieval. Includes a live tool index, equations, dictionary, and research for enterprise AI architects.

Explore the map → GitHub
Case Studies

Problems I solved end to end.

Open SourceShipped
Tokalator: Context Engineering Toolkit
Problem, approach, result
ProblemAI coding assistants burn tokens blindly. No visibility into context cost, no budget guardrails, no cross-model comparison.
ApproachReal-time token budgeting, tab relevance scoring, Cobb-Douglas output-quality models, caching break-even analysis across 17 LLMs.
ResultVS Code extension + web platform shipped. arXiv paper published.
500+ installs -- GitHub stars 17 models MIT license
Next.js·TypeScript·VS Code Extension·Token Economics·MCP
ResearchACM CAIS '26
Multi-Agent Job Classification System
Problem, approach, result
ProblemFree-form job titles in multiple languages can't be reliably mapped to O*NET/ESCO taxonomies at scale.
ApproachMCP-based multi-agent system: planner, search, extraction, and evaluation agents with shared caching layer.
Result95% accuracy on 14K+ positions. 72.5% faster classification. 41.8% fewer API calls. Accepted at ACM.
MCP·LangGraph·Subagent Architecture·AI Search·Caching
EnterpriseShippedAug 2026
ProjectLens: Pricing Internal Tool
Problem, approach, result
ProblemEvery project quote started life as a fresh Excel file. Effort estimation, pricing, proposal drafting, approval and follow-up lived in spreadsheets and inboxes. No shared history, no auditable numbers, no way to ask the data a question.
ApproachBuilt with the Pricing & Revenue team: one tool that runs request intake → effort calculation → pricing → proposal → approval → follow-up end to end. Wired into Microsoft Teams and Azure DevOps so requests arrive where the team already works. Ships a read-only SQL console (SELECT/WITH only, every query in a rolled-back transaction on its own connection) plus a presentation mode that masks figures during screen shares.
Result91% of the manual effort eliminated. 64h of hand work reduced to 5.5h, freeing 59 hours (7.3 working days). 31 Excel files no longer created.
59h freed 91% effort cut 31 Excel files avoided

Savings modelled as 31 pricing runs × (2h − 10min) + 4 intake requests × (30min − 5min). The 2h manual baseline is a team estimate from 2026-07-30, not a measurement.

Applied AI·Microsoft Teams·Azure DevOps·SQL·Workflow Automation
EnterpriseProduction
Enterprise AI Transformation
Problem, approach, result
ProblemManual reporting workflows across Sales, Finance, R&D consuming analyst time with no self-service capability.
ApproachMulti-agent orchestration (LangGraph, MCP, Agno) for text-to-SQL, report automation, and AI feedback loops with human checkpoints.
Result60% reduction in manual workload. 200+ users served. Adopted as group-wide AI standard.
FastAPI·LangGraph·Agno·Pydantic·Arize
Data ProductInfrastructure
300M+ Datapoint Metadata Platform
Problem, approach, result
ProblemNo unified metadata layer across a live job platform. Fragmented schemas, no lineage, analytics bottlenecked by ad-hoc queries.
ApproachDesigned metadata architecture with DataHub governance, real-time pipelines, and product-analytics instrumentation for strategic decisions.
Result300M+ datapoints unified. Became the foundation for all later AI/LLM deployments: embedding generation, RAG retrieval, and context grounding at scale.
DataHub·Azure Synapse·BigQuery·Real-time Pipelines·Retool
Publications

Peer-reviewed and cited.

ACM CAIS '26peer-reviewed
Multi-Agent Position Classification with Tool Orchestration: Use Case System for Occupational Taxonomy Mapping · ACM
alphaxiv discussion
Vahid Faraji et al. · ACM Conference on AI and Agentic Systems, San Jose 2026
arXiv
Tokalator: A Context Engineering Toolkit for AI Coding Assistants · arXiv
Vahid Faraji · arXiv 2604.08290 · April 2026
arXiv
Leveraging LLMs For Turkish Skill Extraction · arXiv
Vahid Faraji et al. · arXiv 2601.22885 · January 2026
Experience

Where I built it.

2025–now

Senior Applied AI Specialist

Kariyer.net (ilab group)
Specs, context and backlog delivery for applied AI transformation across Sales, Finance and R&D. Azure DevOps, Claude Code, Vercel AI SDK, LangChain, Agno, Langfuse.
Google GenAI Leader · Perplexity Fellowship
2022–2025

Product Manager, Data

Kariyer.net
Metadata architecture serving 300M+ datapoints. AI strategy alignment. Pipeline infrastructure enabling later LLM deployments at scale.
2021–2022

Business Analyst

WorqCompany
Forecasting models, dashboards, and process automation for finance and operations.
2019–2021

Co-Founder

Defaro.io
Labour-market analytics startup for compensation intelligence and career pathway analytics. Data pipelines, benchmarking models, product strategy from scratch.
Working Together

Three ways teams bring me in.

Build

Agent systems that survive production.

What that looks like
  • Multi-agent orchestration on MCP, LangGraph or Agno
  • Text-to-SQL and report automation with human checkpoints
  • Evaluation harnesses and observability before launch, not after

Advise

Architecture review before the spend.

What that looks like
  • Context and token strategy: where the budget actually goes
  • Build-versus-buy calls across the agent tooling landscape
  • Data and metadata readiness for RAG and grounding

Train

400+ people trained, 50+ courses.

What that looks like
  • Applied AI for engineering and data teams, 8–18h formats
  • Agents, MCP and tool design, hands-on
  • AI-native development workflow adoption

For Hiring Teams

The short version, in numbers.

  • 9+ years shipping · Istanbul, GMT+3 · remote-friendly
  • 3 papers (1 peer-reviewed ACM) + 1 patent pending
  • Perplexity AI Business Fellowship, programme lead
  • Full CV as PDF · download
Through-line

Twelve years turning messy data into decisions people act on.

2013–15

Master’s, Business Administration

How organisations actually run, and where their decisions get stuck.

2015–18

Research

Measurement and evidence. What separates a defensible number from a convenient one.

2019–21

Defaro.io, co-founder

Built a data product end to end, from raw signal to something customers paid for. Incubated at Özyeğin University.

2019–23

M.A. Economics, Ibn Haldun

Thesis on real-time market analysis. Teaching assistant in econometrics.

2022–25

Kariyer.net, data products

300M+ assets under one metadata architecture. Governance, lineage, real-time pipelines.

2025–

Applied AI

Multi-agent systems that turn free text into structured decisions. Peer-reviewed at ACM CAIS '26.

Twelve years on one problem shape: messy text in, decisions out. The domain keeps changing. The problem doesn’t.
Shipped in Talent Pricing Finance Sales R&D Developer tools Data infrastructure
4 Languages spoken Turkish native, English C1, Spanish at instructor level, Portuguese basic. Japanese in progress, badly.
400+ People taught 50+ courses, an 18-hour applied AI programme, university guest lectures, and career mentoring at the TEV Foundation.
3 Papers, one peer-reviewed ACM member. Writing is how the work gets checked by people who owe me nothing.
0→1 Built unasked Tokalator, awesome-applied-ai and this site all started as problems nobody assigned. The care shows in the small print: where a number is a team estimate rather than a measurement, it says so.
About

The person behind the systems.

Education

03
  • M.A. Economics, Ibn Haldun University
  • Thesis: Real-Time Labor Market Analysis
  • Teaching Asst., Econometrics I

Credentials

08
  • ACM Professional Member
  • Google Generative AI Leader
  • Anthropic: MCP, Agent Skills, Claude API, Claude Code
  • LangChain: LangGraph, LangSmith

Languages

05
  • Turkish: Native
  • English: C1
  • Spanish: Instructor
  • Portuguese: Basic
  • Japanese: Learning

Teaching

03
  • 18h Applied AI Course, Techcareer
  • AI Career Mentor, TEV Foundation
  • Enterprise prompt engineering, Universities

Suggest a time

Pick a slot in your timezone. I'll confirm within 24h.

Istanbul --:-- · You --:--

or book directly on my calendar