Harun Zafer
Data AnalystAI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Software Engineer | AI Consultant | fastsvelte.dev | fastreact.dev | turtledev.io
I help companies automate customer support workflows and document processing using AI-powered chatbots, NLP search, and LLMs. My AI solutions have helped businesses reduce support workloads by 50%, improve response times, and streamline document management. With 15+ years of experience, I’ve worked on AI and NLP projects at Amazon, Tribble AI, and Borealis AI, deploying scalable AI solutions that drive real business impact. 🔹 I specialize in: - AI Chatbots & Customer Support Automation (reducing workload by 50%) - AI for Document Processing & NLP-Powered Search (automating RFPs, contracts, insurance claims) - AI-Powered Internal Knowledge Assistants (enhancing knowledge retrieval for teams) 🔹 What I’m Building: In addition to consulting, I’m building BrokerGuard, an AI-powered risk & fraud detection platform for insurance brokerages. We’re currently in the MVP stage, exploring early adopters and investment opportunities. If you’re interested in collaborating, investing, or learning more, feel free to reach out. 📩 Let’s Talk: If you’re looking to automate customer support, document workflows, or explore AI-powered risk detection, let’s chat! https://calendly.com/harunzafer-dev/15min
Marmara University
Doctor of Philosophy (PhD), Computer Science
January 1, 2013 – Present
Fatih University
Master of Science (MSc), Computer Science
January 1, 2009 – January 1, 2011
Hacettepe University
Bachelor of Science (BSc), Computer Engineering
January 1, 2000 – January 1, 2006
Self Employed
AI & NLP Consultant
February 1, 2025 – Present
Toronto, Ontario, Canada · Remote
BrokerGuard
Founder | BrokerGuard
February 1, 2025 – Present
Tribble
Staff AI Software Engineer
June 1, 2024 – January 1, 2025
Toronto, Ontario, Canada · Remote
Prisma Academy
Board Member
January 1, 2024 – Present
Toronto, Ontario, Canada
Zyfera
AI Software Engineer & Co-Founder
December 1, 2023 – June 1, 2024
Toronto, Ontario, Canada · Remote
Amazon Web Services (AWS)
Software Development Engineer
May 1, 2022 – November 1, 2023
Toronto, Ontario, Canada
KeyNLP
Machine Learning Engineer & Founder
October 1, 2021 – May 1, 2022
Toronto, Ontario, Canada
Amazon
Software Development Engineer
March 1, 2020 – September 1, 2021
Greater Toronto Area, Canada
Borealis AI
Machine Learning Research Developer
January 1, 2019 – March 1, 2020
Greater Toronto Area, Canada
Diply
Senior Software Engineer
June 1, 2018 – August 1, 2018
Greater Toronto Area, Canada
Diligen
Senior Machine Learning Engineer
October 1, 2016 – June 1, 2018
Toronto, ON
TUBITAK
Senior Engineer and Researcher
July 1, 2013 – November 1, 2015
Türkiye
Hecesoft
Startup Founder & Developer
April 1, 2011 – June 1, 2013
Istanbul, Istanbul, Türkiye
Fatih Üniversitesi
Teaching Assistant
September 1, 2009 – June 1, 2013
Istanbul, Istanbul, Türkiye
Turkish Spellcheck Extension for OpenOffice
September 1, 2014 – Present
OpenOffice extension containing comprehensive and up to date dictionaries for Turkish spell-checking. - See more at: http://extensions.openoffice.org/en/project/turkish-spellcheck-dictionary#sthash.bvHqf2GI.dpuf
Turkish Spellcheck Extension for Firefox
September 1, 2014 – Present
A Firefox extension containing comprehensive and up to date dictionaries for Turkish spell-checking.
A fast stemmer for Turkish
August 1, 2014 – Present
Resha is a fast and "less aggressive" stemmer for Turkish written in Java. It uses a stem dictionary which is generated by Nuve using a statistical language model based on morpheme n-grams. So it returns the most possible stem for a word without considering the neighbor words. Main Features Less aggressive and more accurate than the other stemmers for available for Turkish such as the one in SnowBall Contains more than 1.1 million word-stem pairs Based on HashMap, very fast but uses approximately 300 MB of memory. The stemmer class is singleton, thread safe, and lazy initialized
Turkish Search Engine Prototype
December 1, 2013 – January 1, 2015
In a middle-scaled project, following tasks are accomplished: • Developed classifier for genre categorization of web pages by implementing machine learning techniques. Authored an academic conference paper which presents the study. • Developed a Turkish word stemmer to improve both the quality of search and the accuracy of the web page classifier. • Developed a query suggestion module which generates artificial queries due to the lack of an actual query database. Improved the module in order to tolerate the spelling errors made by users. • Performed the setup of the system on a cloud platform and prepared an installation & configuration guide. • Trained teammates and other colleagues about classification, clustering and full text search. • Investigated sentiment analysis for Turkish and introduced a report.
Morphologic Parser for Turkish
January 1, 2012 – Present
Nuve is a Natural Language Processing Library for Turkish. Currently it supports: Morphologic analysis (33K word per second on a i5 2.8 GHz 64 bit machine) Morphologic generation Stemming Sentence (segmentation) boundary detection N-gram extraction Morphologic Analysis is demonstrated here: http://nuvedemo.apphb.com Morphologic Generation is demonstrated here: http://fiilcek.apphb.com More Info: https://github.com/hrzafer/nuve
Fatih Parser
January 1, 2011 – Present
A Generic Syntactic Parser for Turkish and other Turkic Languages. A demo can be seen here: http://fatih-parser-web.herokuapp.com/
Prizma
March 1, 2010 – Present
A Feature Extraction and Selection Tool for Categorizing Text Documents Read directory structured and csv formatted datasets Directory to CSV dataset conversion Support for subcategories Feature Extraction including n-grams terms Best Terms selection based on TF-IDF, Mutual Information, Information Gain, and other metrics Extracted features can be saved in WEKA ARFF format. A more detailed documentation is comming soon...
Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization
Coursera
June 24, 2026 – Present
Structuring Machine Learning Projects
Coursera
June 24, 2026 – Present
Neural Networks and Deep Learning
Coursera
June 24, 2026 – Present
Cultural Fit Analysis
The candidate's diverse experience across startups, large corporations (Amazon, AWS), and research institutions demonstrates adaptability and a broad range of working environments. Their involvement in multiple personal projects and founding companies indicates an entrepreneurial spirit and a passion for innovation. The consistent focus on AI/NLP across various roles aligns well with a data-driven culture. However, the target role is 'Data Analyst' while the candidate's experience is heavily skewed towards 'AI/ML Software Engineer' and 'NLP Consultant'. While there's overlap, a pure Data Analyst role might not fully leverage their deep engineering and ML model development expertise, potentially leading to a mismatch in expectations or underutilization of advanced skills. The candidate's experience is more aligned with a Senior ML Engineer or Data Scientist role.
Soft Skills & Operational Fit
The candidate's extensive project history, including founding startups and leading teams, suggests strong initiative, problem-solving skills, and a proactive approach. Their experience in training colleagues and domain experts indicates good communication and mentorship abilities. The focus on real-world application of AI/ML technologies points to a results-oriented mindset. However, without psychometric test results, a definitive assessment of work attitude, stress handling, and team collaboration is not possible.