cv_v2.tex

%!TEX TS-program = xelatex
%!TEX encoding = UTF-8 Unicode
%
% cv_v2.tex — Mathieu Neron — v2 rewrite
% Template: Awesome-CV (posquit0, LPPL-1.3c)

%-------------------------------------------------------------------------------
% CONFIG
%-------------------------------------------------------------------------------
\documentclass[11pt, letterpaper]{awesome-cv}

\geometry{left=1.4cm, top=.8cm, right=1.4cm, bottom=1.5cm, footskip=.5cm}

\colorlet{awesome}{awesome-skyblue}
\setbool{acvSectionColorHighlight}{true}
\renewcommand{\acvHeaderSocialSep}{\quad\textbar\quad}

%-------------------------------------------------------------------------------
% PERSONAL INFO
%-------------------------------------------------------------------------------
\name{Mathieu}{Néron}
\position{Software Engineer}
\address{San Francisco, CA}
\mobile{(628) 358-8813}
\email{mathieu.neron.ing@gmail.com}
\github{mathieu-neron}
\linkedin{mathieu-néron-b4419522}

%-------------------------------------------------------------------------------
\begin{document}

\makecvheader[C]

%-------------------------------------------------------------------------------
% SUMMARY
%-------------------------------------------------------------------------------
\cvsection{Summary}
\begin{cvparagraph}
Software engineer with 15+ years shipping at scale. Currently building fraud-risk infrastructure at Highnote, processing \textbf{3M+ monthly card transactions} ($\sim$\textbf{\$350M+} in monthly settlement volume) across \textbf{$\sim$45 enterprise programs} and \textbf{1.2M+ cardholders}; previously Senior $\rightarrow$ LMTS at Tableau, building authoring and server features used by \textbf{87K+ customer companies} and \textbf{3M+ Tableau Public creators}. Background spans fintech, data visualization, and e-signature; building applied-ML fluency hands-on (\textbf{Medea}: multi-modal PyTorch classifier with RAG explanation, F1 0.90 on channel-grouped CV).
\end{cvparagraph}

%-------------------------------------------------------------------------------
% EXPERIENCE
%-------------------------------------------------------------------------------
\cvsection{Experience}
\begin{cventries}

  \cventry
    {Senior Software Engineer}
    {Highnote}
    {San Francisco, CA}
    {Aug. 2023 -- Present}
    {
      \begin{cvitems}
        \item {\textit{Java, TypeScript, GraphQL, GCP, Spanner, BigQuery, Temporal, Elasticsearch, Python, ANTLR}}
        \item {Owned \textbf{OFAC sanctions screening} on Highnote's synchronous money-movement critical path (Visa Direct / RTP / ACH); landed a three-layer optimization --- in-memory early-exit, False/True positive lookups deferred to candidate matches, Spanner bounded-staleness reads --- to hit a \textbf{P99 $\leq$ 100\,ms} SLA powering \textbf{Ferry's launch as Highnote's first Instant Payments subscriber} (Dec 2025); \textbf{99.996\%} of $\sim$244K monthly screenings now bypass the database entirely.}
        \item {Designed and shipped a customizable \textbf{geofencing rule} now deployed across \textbf{7 subscribers} (19 rule instances) on Highnote's rule engine; mid-2025 tuning lifted true-positive rate \textbf{3--4$\times$} on the Fleet-program enforcement path.}
        \item {Built the \textbf{real-time velocity feature store} powering Highnote's fraud rule engine --- intra-hour merchant/BIN/card auth-velocity counters (count + amount) written on the synchronous auth path and served to rule evaluation via a CQRS read API; collapsed the Redis read from N round-trips to \textbf{2 concurrent MGETs ($\sim$1 RTT)}, engineered to hold \textbf{P95 $<$ 100\,ms under load} as the program scales.}
        \item {Led migration of the core SMS stack from AWS Pinpoint to \textbf{Twilio} and built the platform's first \textbf{bi-directional SMS fraud-detection workflow} on \textbf{Temporal} --- conversation state lives in Temporal's durable workflows, so the SMS layer uses Twilio's lightweight Messaging API rather than the heavier and pricier Conversations API; anchors a flagship enterprise rollout.}
        \item {Drove adoption of \textbf{Temporal} (durable workflows) and \textbf{Elasticsearch} as first production use cases at Highnote; both now standard across multiple teams.}
        \item {Extended the \textbf{ANTLR grammar} powering Highnote's natural-language query layer with fuzzy / typo-tolerant search.}
      \end{cvitems}
    }

  \cventry
    {Senior Software Engineer $\rightarrow$ LMTS}
    {Tableau (Salesforce)}
    {Palo Alto, CA}
    {Jun. 2019 -- Jul. 2023}
    {
      \begin{cvitems}
        \item {\textit{React, Redux, TypeScript, Java, C++, Qt} \quad---\quad Promoted to Lead Member of Technical Staff in 2020.}
        \item {Onboarded a new AI-predictive-model content type in Tableau Server for \textbf{Salesforce Einstein Discovery} integration --- contributing to the server stack delivering \textbf{80B+ AI-powered predictions per day}.}
        \item {Drove the \textbf{Object Graph} view in Tableau's Data pane from beta to GA --- the default relationships editor in Tableau's multi-table data model, part of the authoring stack behind Tableau Public's \textbf{7M+ vizzes} from \textbf{3M+ creators}; shipped transition animations, tooltips, sorting, and re-rooting.}
        \item {Contributed to the \textbf{Data Preview Area} hybridization --- metadata grid plus embedded relationship editor in a single surface.}
      \end{cvitems}
    }

  \cventry
    {Java Developer}
    {Paysafe}
    {Montréal, Canada}
    {2016 -- 2019}
    {
      \begin{cvitems}
        \item {\textit{Angular 2+, TypeScript, Java, Spring, Oracle}}
        \item {Shipped the new Angular 2+ \textbf{Merchant Portal} frontend from greenfield.}
        \item {\textbf{Cut frontend test runtime 6$\times$} by migrating from PhantomJS to Nightmare.}
      \end{cvitems}
    }

  \cventry
    {Java Developer}
    {eSignLive}
    {Montréal, Canada}
    {2014 -- 2016}
    {
      \begin{cvitems}
        \item {\textit{Java, Spring, Guava, AWS, MySQL, Jenkins}}
        \item {\textbf{Cut Windows document converter downtime by 50\%}.}
        \item {Integrated Equifax, Twilio, PDFBox, and Google Noto Fonts.}
      \end{cvitems}
    }

  \cventry
    {Application Developer}
    {Avanade}
    {Montréal, Canada}
    {2012 -- 2014}
    {
      \begin{cvitems}
        \item {Enterprise web engineering for Canadian National Railway (\textit{www.cn.ca} redesign on Sitecore / ASP.NET) and Hydro-Québec (internal time-tracking app; led frontend-stack selection --- SAPUI5 + Infragistics).}
      \end{cvitems}
    }

\end{cventries}

\newpage

%-------------------------------------------------------------------------------
% PROJECTS
%-------------------------------------------------------------------------------
\cvsection{Projects}
\begin{cventries}

  \cventry
    {RealTube --- crowdsourced AI-generated YouTube content filter}
    {\href{https://github.com/mathieu-neron/RealTube}{github.com/mathieu-neron/RealTube}}
    {Personal learning project}
    {2026}
    {
      \begin{cvitems}
        \item {Full-stack browser extension (Chrome/Firefox, Manifest V3) for crowdsourced flagging and hiding of AI-generated YouTube content across feed, search, recommendations, and Shorts.}
        \item {Built as a hands-on Go + Python learning exercise: \textbf{Go} (Fiber v3) on the low-latency lookup path, \textbf{Python + FastAPI} for the inference-adjacent services, \textbf{TypeScript + React 19} extension, PostgreSQL + Redis + Docker + Prometheus.}
        \item {Privacy-engineered URL lookup using a \textbf{k-anonymity hash-prefix} scheme (HIBP-style) so the backend never sees full URLs in plaintext; contract-tested API and end-to-end test coverage.}
      \end{cvitems}
    }

  \cventry
    {Medea --- multi-modal AI-generated YouTube classifier}
    {\href{https://github.com/mathieu-neron/medea}{github.com/mathieu-neron/medea}}
    {Personal learning project (natural follow-up to RealTube)}
    {2026}
    {
      \begin{cvitems}
        \item {Built an end-to-end ML pipeline to learn the full applied-ML stack hands-on: yt-dlp ingest $\rightarrow$ four-modality feature extraction (\textbf{CLIP ViT-B/32} visual, \textbf{faster-whisper} + \textbf{wav2vec2} anti-spoofing audio, \textbf{MiniLM} text, handcrafted scalars) $\rightarrow$ \textbf{Chroma} vector DB $\rightarrow$ small \textbf{PyTorch} MLP head $\rightarrow$ FastAPI server with retrieval-augmented \textbf{local-LLM rationale} via Ollama.}
        \item {Used \textbf{channel-grouped LeaveOneGroupOut CV} (10 folds, $\sim$94-clip dataset) after UMAP showed channels as 10 distinct clusters --- random video-level splits would have memorized channel identity, not ``AI-ness.'' Final MLP \textbf{F1 0.90} / \textbf{ROC-AUC 0.96} on this honest split.}
        \item {Diagnosed a silent-audio failure mode (anti-spoof model misfires on no-speech clips) and fixed it with \texttt{has\_speech} $\times$ \texttt{ai\_voice\_prob} interaction features (F1 0.87 $\rightarrow$ 0.90); added a \textbf{temporal prior cap} for pre-2017 uploads to bound MLP extrapolation outside the training cone.}
      \end{cvitems}
    }

\end{cventries}

%-------------------------------------------------------------------------------
% SKILLS
%-------------------------------------------------------------------------------
\cvsection{Skills}
\begin{cvskills}
  \cvskill
    {Languages}
    {Java, TypeScript/JavaScript, Python, Go, C/C++, C\#}
  \cvskill
    {Frameworks}
    {Spring, Temporal, Node.js, React/Redux, Angular, GraphQL, REST, Qt, .NET, ANTLR}
  \cvskill
    {Platforms}
    {GCP, AWS, Spanner, BigQuery, PostgreSQL, Elasticsearch, Redis, Oracle, MySQL, Docker}
  \cvskill
    {ML / AI}
    {PyTorch, Hugging Face Transformers, sentence-transformers, CLIP, Whisper, Chroma, MLflow, Ollama, RAG}
  \cvskill
    {Spoken}
    {French (native), English (fluent)}
\end{cvskills}

%-------------------------------------------------------------------------------
% EDUCATION
%-------------------------------------------------------------------------------
\cvsection{Education}
\begin{cventries}

  \cventry
    {B.Eng., Computer Engineering}
    {Université de Sherbrooke}
    {Sherbrooke, Canada}
    {2007 -- 2011}
    {
      \begin{cvitems}
        \item {Specialization in Software Engineering and Artificial Intelligence.}
      \end{cvitems}
    }

\end{cventries}

%-------------------------------------------------------------------------------
% EARLY WORK
%-------------------------------------------------------------------------------
\cvsection{Internships}
\begin{cventries}

  \cventry
    {Software Engineer Intern}
    {Electronic Arts Canada}
    {Burnaby, Canada}
    {Summer 2011}
    {
      \begin{cvitems}
        \item {Built a distributed C++ / C\# load-test harness running thousands of concurrent bots against \textbf{SSX}'s multiplayer server; used internally and demoed at \textbf{Gamescom 2011}.}
      \end{cvitems}
    }

  \cventry
    {Earlier Internships}
    {Autodesk \quad\textbar\quad Ubisoft \quad\textbar\quad Lockheed Martin}
    {Canada}
    {2008 -- 2010}
    {
      \begin{cvitems}
        \item {\textbf{Autodesk} (2010) --- animation plug-in inside Maya (C++ / Qt). \textbf{Ubisoft} (2009) --- QA automation (C\# / .NET). \textbf{Lockheed Martin} (2008) --- software evaluation.}
      \end{cvitems}
    }

\end{cventries}

\end{document}