Unit 14 — Career Pathways & Professional Development

CYB-4203/6203: Secure and Trustworthy AI

Wednesday, April 29, 2026

Dallas Elleman — Spring 2026

Course Orientation

Section 4 — SYNTHESIS — Unit 14

Last session — Pres 21

Risk, Audit & Industry Landscape

Where the field is and where it's going

→

Today — Pres 22

Career Pathways & Professional Development

Where you go from here

Plan: Final-project status → 14.1 careers (deep) → 14.2 pro dev → 14.3 synthesis → workshop & close

Final Project — Teams & Status

Teams have been selected. Build is almost complete — assignment description releases later today.
Project due Wed May 13, 11:59 PM.

Team 1

Nathan · Teresa · Ahmad

Team 2

Benton · Gangalaxmi · Molly

Team 3

Leyla · Jasmine · Oliver

Team 4

Mark · Talha · Christian

Team 5

Phillip · Xingzhou · Calvin

Team 6

Sarah · Helaina

14.1 — Career Pathways & Certifications

Top student-voted topic in Unit 14 — deeper treatment

14.1A — Roles

Eight job titles you'll see in 2026 postings. None existed as a labeled role five years ago; most are still settling.

AI Security Engineer

Hardens AI in production — guardrails, sandboxing, output filtering, tool-call governance. Infosec + ML/AI.

AI Red-Teamer

Adversarial probing of models, agents, pipelines. Prompt injection, jailbreaks, tool abuse, supply-chain. What the final project simulates.

AI Safety Researcher

Frontier-lab + AISI work on alignment, evals, capability surveying, deceptive-behavior research. PhD or research-engineer track.

Evals / Research Engineer

Designs and runs benchmarks for capability and safety. METR, Apollo, Pattern Labs, MLCommons, lab-internal eval teams.

ML Platform / Infra

Training, serving, MLOps stack. Supply-chain hardening, model registry, deployment pipelines — the platform side of secure AI.

AI Risk & Governance

NIST AI RMF, EU AI Act, sector regs. Risk committees, model-risk mgmt (SR 11-7), audit prep, ethics-board work.

Policy / Standards

CAISI, AISIs, OECD, EU AI Office, congressional/executive staff, think tanks (CSET, RAND). Where the rules get written.

Applied Alignment / Interpretability

Mech interp, scalable oversight, constitutional AI, interpretability for safety. Most lab "research scientist" postings.

CISA NICCS Career Map · AISafety.com jobs · 80,000 Hours

14.1B — Where the Jobs Actually Live

Global / National

Frontier labs: Anthropic Frontier Red Team, OpenAI Frontier Evals, DeepMind Frontier Safety ($136–245K), Meta Security, xAI
AI Safety Institutes: UK AISI (£65–145K, sponsors visas), US AISI / NIST CAISI, Japan AISI, Singapore AI Safety Hub, EU AI Office
Eval & safety orgs: METR, Apollo, Pattern Labs, MLCommons, FAR.AI, CAIS
U.S. federal: CISA (~329 critical hires, ~1K vacancies), NIST CAISI on USAJobs, NSA, sector regulators
Big-tech product side: Microsoft AI Red Team, Google Trust & Safety, AWS Responsible AI, Meta Purple Llama

Regional & Local — Tulsa / Oklahoma

THETA — $51M EDA Tech Hub; cyber + GenAI + autonomy
OCII — $75M projected; runs OK's first cyber range
Local employers: Williams, BOK Financial, Spirit AeroSystems, ONEOK, Targa, QuikTrip, Deloitte (Tulsa), Kaiser-Francis, PSO/AEP
Tulsa startups (TIL): Asemio, True Digital Security, Verinovum
Federal in OK: Tinker AFB cyber (OKC), FAA Mike Monroney Aeronautical Center
Regional hubs: Plano-Dallas + Austin (NSA Texas), Kansas City, Colorado Springs (Space Force/AFRL)

Tulsa cyber employers — Nucamp · EDA Tulsa Tech Hub · CISA hiring plan

14.1C — How to Actually Find These Jobs

Job boards, student-only pipelines, and federal programs that give you tuition + a job at graduation.

General-purpose boards (start here)

Handshake (built for students; school login filters) · LinkedIn Jobs (largest base + networking) · USAJobs (only path to most fed cyber roles) · Wellfound (small/fast AI labs & sec startups) · Indeed

Specialized boards & lists

speedyapply 2026 AI/ML Jobs (GitHub, daily) · AISafety.com Jobs (curated AI safety/security) · CyberSeek (NICE/NIST regional heatmap) · ClearanceJobs (defense/intel; many sponsor your clearance) · Built In

CyberAI SFS — tuition + stipend + guaranteed job

NSF CyberAI Corps Scholarship for Service — formerly CyberCorps SFS, renamed in 2026 to explicitly cover AI security. Full tuition + $27K/yr (UG) or $37K/yr (grad) + $6K/yr pro-dev. Service obligation = years funded. Apply through your school.

Other federal student pathways

CISA Student Internships (100 dedicated SFS summer slots in 2026) · NSA Stokes (full tuition + post-grad job) · DoD SMART (up to $46K/yr stipend + DoD service) · Pathways Program

Most summer 2026 internships closed in fall 2025; for summer 2027 start applying Aug–Sep 2026. Federal roles take 4–6 months — apply earlier.

14.1D — The Numbers (and Who's Tracking)

+33%

Info Security Analyst growth, 2024–2034
BLS OOH

~17.3K

Annual openings (info-sec analyst), 2024–34
BLS Employment Projections

+88%

AI/ML hiring YoY growth in 2025
Ravio 2026 trends

$206K

Avg AI engineer salary (2025), +$50K YoY
KORE1 AI Jobs 2026

+56%

Pay premium for workers with AI skills
StationX market stats

87%

Cyber pros: AI vulns are the fastest-growing risk (2025)
Cybersecurity Jobs 2026

62%

Orgs citing security/risk as top agentic-AI barrier
Stanford 2026 AI Index

~10%

Cyber job postings now referencing AI skills
StationX market stats

~329

CISA "mission-critical" hires planned (~1K vacancies)
Federal News Network

Who tracks this stuff (bookmark these)

BLS OOH · BLS MLR — AI impacts · Stanford HAI AI Index · ISC2 Workforce Study · WEF Future of Jobs · Lightcast

The macro picture: classical info-sec is one of the fastest-growing occupations BLS tracks (+33% over a decade). On top of that, AI-specific roles are growing dramatically faster (+88% YoY in 2025) and pay a steep premium for the AI skill overlay (+56%). That premium is the empirical answer to "is this course worth it" — the AI security overlay on a classical security baseline is the expensive, scarce skill in this market. The 62% Stanford-AI-Index number is the one to carry around: most enterprises trying to scale agentic AI cite security and risk as the dominant blocker. That is the demand signal. The students in this room are training to fill exactly that gap. The CISA number is the closest read on federal demand: ~1,000 vacancies, 329 "mission-critical" hires being prioritized. Federal pipeline is messy in 2026 but the seats are real. Bookmarks bar: BLS OOH, Stanford AI Index, ISC2, WEF Future of Jobs, Lightcast. These are the reputable trackers; everything else is a re-package.

14.1E — Certifications, an Honest Read

Be strategic — certs are expensive. Recommendations reflect what 2026 entry-level postings actually request.

Entry-level priority — do this one first

CompTIA Security+ (~$404) — appears in ~70% of entry-level postings; DoD 8140 baseline. Pair with Google Cybersecurity Cert (~$59/mo) for hands-on Linux/Python/SIEM. Then Network+ if your networking is weak.

Mid-level & specialty

CompTIA CySA+ (~$404; SOC roles) · OSCP / PEN-200 (~$1,600+; offensive cred after 1–2 yrs exp) · CISSP (mgmt) · GIAC (specialty depth).

Cloud + AI-specific (the new layer)

CompTIA SecAI+ — NEW in 2026, first major vendor-neutral AI security cert (adv ML, prompt injection, governance). AWS Sec Specialty / Azure Sec Eng (~$300 each). IAPP AIGP for governance/GRC track.

What actually beats certs — a portfolio

Final-project red-team writeup. DEF CON AI Village placements. A working eval or model-card critique. Active GitHub. HackerOne / Bugcrowd findings. Skip CAISP/CAISS — unproven in entry-level hiring.

CISA NICCS — full cert framework · AI/cyber Master's comparison · ISACA AAISM (mid-career)

Honest take. Classical certs (CISSP, OSCP, GIAC, Security+) are well-understood; hiring managers can map them to expected skills. The new AI-specific certs (AAISM, AIGP, AICM) launched in the last 12–24 months — they signal that you are tracking the field but don't yet have the labor-market pricing the older certs do. For AI-specific roles — red-team, evals, applied alignment — a portfolio dominates a cert. Public CTF placements, a written-up red-team exercise (the final project counts), a critique of a model card, an eval you built. Signal > credential here. For governance, policy, and traditional infosec roles, the certs still pull weight. Practical advice: undergrads — Sec+ first, then OSCP if offensive, then layer one AI-specific cert. Working professionals — CISSP/CISM if the goal is mgmt; AIGP if the goal is governance.

14.1F — Competitions That Get You Hired + Safety Fellowships

Hiring managers cite CTFs as a tiebreaker. Safety fellowships welcome cyber backgrounds — ML research experience not required.

Cyber competitions — pick one this semester

picoCTF (CMU; beginner; free, world's largest) · NCL (~$35; published Scouting Report on resume) · CCDC (blue-team; SOC-pipeline) · CSAW (NYU; $1M+ in prizes) · MITRE eCTF · US Cyber Games

Hands-on platforms (always-on labs)

Hack The Box · TryHackMe — not competitions per se but a high HTB rank is real currency on a resume. For ML side: Kaggle bronze = baseline competence; expert = interview-worthy alone. MLSafety / Trojan Detection challenges directly map to this course.

AI safety fellowships — the AI-side ladder

Anthropic Fellows (~$3,850/wk; cyber backgrounds explicitly welcome; 40%+ join Anthropic full-time) · MATS (10-wk paid Berkeley fellowship; leading alignment-research entry point) · CHAI (UC Berkeley; $5K/mo in-person)

Lower-barrier safety on-ramps

CAIS Research Engineer Internship (SF summer; adversarial robustness, trojans, OOD detection — direct course overlap) · AI Safety Camp (free, remote, project-based) · BlueDot Impact AISF (free 8–12wk curriculum; technical or governance track)

DEF CON AI Village for the adversarial-ML CTF that lives at the intersection of both columns — the single most-relevant competition for this course.

14.2 — Professional Development

Communities, conferences, and the "keep going to school" question

14.2A — Communities, Conferences, Reading Habits

Communities, working groups & meetups

OWASP GenAI · MITRE ATLAS WG · AI Village (DEF CON) · Alignment Forum / LessWrong · EleutherAI Discord · DEF CON Groups · BSides ($20–40 student) · ISC2 Student ($20/yr) · local: Tulsa Tech Fellows, OKC InfraGard

Conferences worth a budget line

DEF CON AI Village · USENIX Security · NeurIPS / ICLR safety workshops · RSA · AAAI / AIES · SafeAI Workshop · Black Hat AI tracks

Reading list — primary sources

Anthropic, OpenAI, DeepMind, Apollo, METR · UK AISI + US AISI · transformer-circuits.pub for interpretability

Habits + experience without a job title

AI Incident DB weekly · 1 frontier-lab paper/wk · ATLAS diff-watch · OWASP LLM Top 10 · NVD. Build signal: active GitHub, blog posts, HackerOne / Bugcrowd, home lab (Wazuh + Proxmox). Volunteer for nonprofits / EFF / school IT.

AISafety.com aggregates jobs, reading lists, and program directories — the single best starting bookmark.

14.2B — AI/Security Grad Programs

CMU

MSAIE-IS — AI Eng + Info Security.

Stanford

AI Safety Center + AI Grad Cert.

Georgia Tech

OMSCS Computing Systems — spec + ML electives.

Purdue

Applied AI + Cyber cert → new MS 2026.

USC

MS Cyber Sec Engineering — Viterbi.

Old Dominion

MS Cyber — AI Security concentration.

FIU

MS CompEng Sec — AI4Cyber + Cyber4AI, ~10mo online.

OU Polytechnic

BS Applied AI — in-state counterpart.

The University of Tulsa — the in-house pipeline

Brand-new: BS Applied AI (just launched; double-major friendly with CS/eng/cyber). · Established: MS Cyber Security (Online) — #23 nationally; founded 2001, one of the first in the country. · MS Pro Track (hybrid online + on-campus immersives). · PhD Cyber Studies — full BS → MS → PhD pipeline. · NSA + DHS CAE in Research & Cyber Defense Ed. · Lead institution on OCII + the $51M THETA award.

AI/Cyber Master's comparison · TU AI degree launch

The honest "should I keep going to school" answer depends on which lane: research (alignment, interp, evals) realistically wants a PhD or a research-engineer track that mimics one. Practitioner lanes (red-team, AI security engineering, governance) are well-served by an MS or by accumulated portfolio + classical certs. The TU plug is real, not a courtesy. The Applied AI BS is brand-new and is the in-house counterpart for undergrads sitting in this room. The MS Cyber Security has a 25-year legacy — founded in 2001, one of the earliest in the country, ranked #23 nationally. The PhD pipeline closes the loop. And TU is the lead on OCII + THETA, which means the research and the local labor market are tightly coupled. For people in this room: if you're an undergrad, talk to your advisor about whether the Applied AI BS makes sense as a double-major with what you're already doing. If you're an MS-track student, the Professional Track is built for you. If you want the research path, the PhD pipeline through Cyber Studies is the path to OCII research roles.

14.3 — Course Synthesis

The four-arc map — what to take with you

14.3 — The Course in One Slide

Section 1 — WHY

Ethics, harms, regulation. Frames why secure & trustworthy AI is a load-bearing field, not a compliance afterthought.

Section 2 — WHAT

Threat landscape, ML + LLM attack vectors, privacy & bias. The technical surface that has to be defended.

Section 3 — HOW

PETs, testing & red-teaming, MLOps + guardrails, risk mgmt, audit & disclosure. The operational toolkit.

Section 4 — SYNTHESIS

Industry landscape, careers, your final project. The integrative artifact that ties WHY/WHAT/HOW together.

What I want you to leave with: an instinct for the layered-defense story, a working threat-modeling reflex, and a portfolio piece — the final project — you can point to.

Your Next 30 Days — A Concrete Plan

From the class handout. The work compounds — do the things that put your name in front of recruiters.

Week 1 — presence

Create accounts on Handshake, LinkedIn, and TryHackMe or HTB. Polish LinkedIn (real photo, real summary). Update resume with one specific target role in mind.

Week 2 — skills in motion

Start one CTF or hands-on platform. Begin Security+ study or the Google Cybersecurity Cert. Identify three local meetups or BSides events to attend this semester.

Week 3 — programs & people

Look up your school's SFS contact (or the nearest one). Apply to NCL for next semester. Reach out to one alum or LinkedIn contact in your target role for an informational chat.

Week 4 — ship & apply

Push a first project to GitHub: a CTF writeup, a security tool, a reproduction of a paper from this course (your final-project red-team writeup counts). Apply to 5 internships, even if it feels too early.

Strategic notes from the handout: pick a lane within 6 months (SOC, offense, GRC, cloud, AI/ML security, appsec). AI fluency is now a hiring differentiator even for traditional cyber roles. Apply earlier than you think — federal pathways take 4–6 months. Three informational interviews per month is the most underused career tool.

Final-Project Workshop Time

Open studio — team check-ins, blockers, infrastructure questions

What's Next

Monday, May 4 — Final Class Session

Course synthesis & wrap-up + final exam review. (No final-project presentations on this date.)

Final Project

Due Wednesday, May 13 at 11:59 PM. Workshop time today; check Slack for any updates.

Final Exam Time — Help Us Pick a Slot

Scan to submit when your other finals are scheduled. Once everyone's other exams are in, I'll pick a 2h25m window that doesn't conflict.