(BERKELEY, CALIFORNIA) Computer science once meant near‑certain hiring for new grads. Not anymore, says Hany Farid, a professor at UC Berkeley, who warned this fall that the job market has shifted sharply—even for students from top programs. Speaking on Nova’s “Particles of Thought” podcast, he described a reset in how tech companies hire and what skills they prize. That reset has immediate stakes for international students in the United States 🇺🇸 whose ability to stay and work often depends on securing timely employment to support Optional Practical Training (OPT), STEM OPT, and later H‑1B or employment‑based green cards.
Farid recalled that Berkeley graduates “had the run of the place” just a few years ago, often choosing among multiple offers. Now, many feel lucky to get one. He cited the struggle of astrophysicist Hakeem Oluseyi’s son—finishing a CS degree but finding it hard to land a role—as a window into a broader trend across elite programs.

The reasons are layered: AI and automation changing how code gets written, companies rethinking headcount, a thinner set of traditional roles, and more graduates chasing fewer “safe” openings. Those pressures carry extra weight for international students. Deadlines tied to work authorization add a clock that domestic students don’t face.
How work authorization timing interacts with the market
Under current rules, F‑1 students usually apply for post‑completion OPT with Form I-765 (Application for Employment Authorization). STEM degree holders can seek a 24‑month extension using a formal training plan (Form I‑983) with a qualifying E‑Verify employer. When hiring slows or roles shift toward niche skills, students risk:
- Missing OPT start windows
- Losing F‑1 status
- Running out of time before the annual H‑1B selection
Policy context and staffing trends
Farid’s comments come as tech layoffs and team reshaping continue. Industry counts show:
- >264,000 tech layoffs in 2023
- 152,000 in 2024
- Continued cuts into 2025, with many reductions hitting junior and mid‑level roles
Analysis by VisaVerge.com finds employers have kept or grown teams in areas like AI, data science, cybersecurity, and cloud infrastructure—fields where specialized skills and real‑world project work matter most.
The Bureau of Labor Statistics projects about 356,700 openings per year through 2033 across computer and information technology jobs, but the type of demand is shifting. Farid urges students to change course: don’t aim only for narrow mastery—learn across domains and use AI tools rather than trying to compete with them. In short: breadth + proof you can ship real work matters more than ever.
Important timelines for immigrant students:
– OPT and STEM OPT are time‑bound and require paid, qualifying work tied to the degree field.
– H‑1B selection is annual and random; employers file Form I‑129 after registration.
– Employment‑based green cards depend on employer sponsorship, often via Form I‑140, and Form I‑485 when a visa number is current.
In a tight market, each missed interview or rescinded offer can create status risks—especially for students nearing the end of their OPT period.
Impact on international graduates and employers
Farid’s guidance reflects several converging forces:
- AI tools that write or check code reduce the need for large teams of generalists.
- Companies recalibrate how many developers they need and what stacks matter.
- A surge of graduates with similar skills increases competition for entry‑level roles.
- Hiring screens now prioritize depth in AI integration, security, data, and cross‑disciplinary fluency, plus resilience and fast learning.
For F‑1 students, the immediate effects show up in job search timelines. Many used to secure roles months before graduation; now some are still interviewing as OPT windows approach.
Practical consequences and employer expectations:
- Students can apply for post‑completion OPT up to 90 days before program end; processing can take months.
- Filing Form I‑765 early, with a clear plan for start dates, reduces the risk of gaps.
- For STEM OPT, the training plan on Form I‑983 must align with the degree and list specific learning goals under a qualified employer.
- Employers that have cut generalist roles still hire for targeted needs. Workers who use AI effectively will outpace those who don’t.
In immigration terms, international candidates should demonstrate safe and effective AI use—code generation, testing, data workflows, model integration—while managing confidentiality and policy concerns. A portfolio blending software engineering with AI, security, cloud, or domain knowledge (healthcare, finance, climate) helps both hiring managers and immigration counsel when explaining the specialty nature of a role for H‑1B or green card petitions.
The layoffs also change risk management around H‑1B season. If an offer falls through after H‑1B registration but before filing, sponsorship may stall for the year. Students on STEM OPT should keep backup plans:
- Multiple applications
- Short contract roles that meet program rules
- Ongoing skill upgrades aligned with market needs
Farid’s warning—that what felt “future‑proof” changed in four years—has personal stakes for families. Some students may accept later offers to preserve OPT and a path to H‑1B; others may pivot quickly to AI‑heavy bootcamps, open‑source contributions, or domain internships that count toward STEM OPT while building specialized proof.
Practical steps and key documents
Farid’s advice to “be polymathic” can feel abstract. Here are concrete immigration and hiring actions international students can take now.
Actionable career and portfolio steps
- Build adjacent skills you can show: data pipelines, MLOps basics, human–computer interaction, security concepts. Tie each to a small, working project.
- Use AI in your work and document it: how you prompt, review outputs, handle bias, and secure data.
- Assemble a hybrid portfolio: combine software builds with domain use cases (healthcare triage, payments risk, grid forecasting, climate analytics).
- Stay visible: contribute to open‑source, present at meetups, and write short posts explaining your design choices.
- Track immigration timing: file OPT early, set realistic start dates, and keep records ready for onboarding and E‑Verify if seeking STEM OPT.
When paperwork matters, accuracy and timing are critical. Key forms and resources:
- Form I‑765 (Application for Employment Authorization) — file for OPT/STEM OPT: Link: USCIS Form I-765
- Form I‑983 (STEM OPT Training Plan) — coordinate with your DSO and employer: Link: Form I‑983 Overview
- Form I‑129 — employer files for H‑1B after registration: Link: USCIS Form I-129
- Form I‑140 — employment‑based green card petition: Link: USCIS Form I-140
- Form I‑485 — adjustment of status when a visa number is available: Link: USCIS Form I-485
For a clear overview of student work options, see USCIS on OPT, STEM OPT, and reporting duties: Optional Practical Training (OPT).
What this means for 2025 graduates
The market remains active but is tilted toward specialties. Demand looks strongest in AI, cybersecurity, and cloud engineering, while entry‑level generalist roles face stiffer competition. BLS projections suggest long‑term openings, so the near‑term picture is tough but not closed.
Strategy checklist for students
- Show depth in a focused area and prove you can use AI well.
- Keep learning visible with projects, open‑source contributions, and measurable outcomes.
- File OPT/STEM OPT paperwork early and accurately.
- Maintain a financial buffer for slower hiring cycles.
- Consider short contracts that comply with STEM OPT rules while building proof of skill.
- Document achievements with metrics, code links, and concise write‑ups that hiring managers can review quickly.
Farid’s comparison: “Lawyers who use AI will put those who don’t out of business.” The same applies in tech hiring—and in immigration filings. A portfolio that clearly shows AI use, data care, and responsible deployment helps an employer justify an H‑1B “specialty occupation” and, later, an I‑140 claiming advanced duties.
VisaVerge.com reports signs of recovery could arrive as startups scale through late 2025, but specialists are likely to benefit first. That aligns with Farid’s push for breadth plus a sharp edge: not a random list of tools, but a clear combo—e.g., backend engineering + MLOps for small language models; frontend engineering + AI‑driven UX testing; or data engineering + security for healthcare analytics.
Farid’s bottom line: the industry changed fast. For UC Berkeley students and peers nationwide, the next move is to change how they prepare—wider skill sets, real projects, and smart use of AI. For international students, add one more rule: pair that strategy with early filings, careful status planning, and backup options. In this job market, the strongest path is not only to code well, but to show why your work matters—and to have the right documents ready when the offer finally lands.
This Article in a Nutshell
UC Berkeley professor Hany Farid warns the computer science job market has shifted sharply, reducing the nearly certain hiring that once favored new graduates. Contributing factors include AI and automation, company headcount adjustments, and a larger pool of similarly skilled graduates—resulting in fewer generalist entry roles and stronger demand for specialists in AI, cybersecurity, data, and cloud infrastructure. International students face added vulnerabilities because OPT and STEM OPT are time‑bound and H‑1B selection is annual and random. Farid advises students to develop adjacent skills, document AI use, deliver real projects, and file OPT/STEM paperwork early. Employers still hire but prioritize demonstrable specialization, cross‑disciplinary fluency, and proven ability to ship work; specialists are likely to benefit first as recovery occurs late 2025.