Skip to content

How to Negotiate Your AI/ML Compensation: Scripts That Actually Work

DS
LDS Team
Let's Data Science
17 minAudio
Listen Along
0:00/ 0:00
AI voice

Signing bonuses at Meta for ML engineers run $30,000 to $80,000 for mid-level roles, with senior engineers regularly clearing $80,000 to $150,000 in signing alone. At Google, comparable ranges apply — money most candidates never see because they didn't ask.

According to CareerBuilder research, 73% of employers are willing to negotiate an initial offer — yet 55% of candidates never ask for more.

This article is for the person holding an offer right now. Not someday. Right now. You have more negotiating power than you think, a short window to use it, and scripts in this article you can copy and send today.

The Fear Keeping Money in Your Recruiter's Budget

The most common reason candidates don't negotiate is fear of rescission — the terrifying possibility that asking for more money gets the offer pulled. This fear is almost entirely unfounded.

Recruiters invest weeks or months closing a candidate. Rescinding an offer because someone asked for a higher salary is extraordinarily rare and typically only happens when a candidate negotiates in bad faith: inventing a competing offer that doesn't exist, becoming hostile, or making demands that are genuinely insulting relative to the role's band. A polite, professional counter costs you nothing.

The Salary.com figure that gets cited is 19% of candidates "losing an offer after negotiating" — but that stat conflates rescission with withdrawal. Many of those candidates chose to decline after the company wouldn't budge. Offers are very rarely rescinded simply because a candidate asked for more. Employers know negotiation is normal. Recruiters budget for it.

Key Insight: If a company rescinds your offer because you professionally and politely asked for more compensation, that company almost certainly had deeper cultural problems you would have discovered after joining. The rescission did you a favor.

The Anatomy of an AI/ML Compensation Package

Before you can negotiate effectively, you need to understand what's actually on the table. Most candidates focus entirely on base salary, which is often the least flexible part of an offer.

ComponentTypical Range (Senior ML Engineer)Flexibility
Base salary$160K to $230KLow to medium (band-constrained)
Annual bonus10% to 25% of baseLow (formula-driven)
Signing bonus$20K to $150K+High
Equity (RSUs)$80K to $400K over 4 yearsHigh
Refresh grants25% to 75% of initial grant annuallyNegotiable at offer
Remote stipend$0 to $5K/yearMedium

Source: Levels.fyi ML/AI Engineer data, Q1 2026; TeamRora AI/ML Compensation Guide, 2025

Base salary is often capped by compensation bands that require VP-level approval to exceed. Signing bonuses and equity grants have much more flexibility because they involve a single budget decision rather than ongoing payroll commitments. When a recruiter says "we can't go higher on base," that is often genuinely true — but it doesn't mean the conversation is over.

Pro Tip: Treat every compensation component as a separate negotiation. If base is capped, immediately shift to signing. If signing has a ceiling, push on equity quantity or accelerated vesting. These aren't consolation prizes; they are often larger than the base difference you were trying to close.

Knowing Your Number Before You Talk

Walking into a negotiation without a clear target number is the fastest way to leave money behind. Here's how to build your number with current market data.

Levels.fyi is the most credible source for tech compensation. It's crowd-sourced from verified offer letters and includes base, bonus, equity, and signing broken out separately. For any FAANG or Tier-1 tech negotiation, build your anchor from here. The median ML/AI software engineer total compensation sits at $243,333 as of Q1 2026 (Levels.fyi). Senior ML engineers at major tech companies go well beyond that: the Levels.fyi median for ML engineers at Meta is $411,940, at Google it is $313,225, and at Snap it is $530,000 — all verified as of Q1 2026.

State salary transparency laws are a direct line to the actual band before you even apply. As of 2026, 14 states and Washington D.C. require employers to disclose pay ranges in job postings. Colorado, California, Washington, New York, Illinois, New Jersey, Massachusetts, Vermont, Minnesota, Nevada, Rhode Island, Connecticut, Hawaii, and Maryland all have active requirements, with most requiring disclosure in job postings directly. If you're applying in any of these states, the range is in the posting. That floor is your starting point, not your ceiling.

PwC's 2025 Global AI Jobs Barometer provides a compelling citation for live negotiations: workers with AI skills command a 56% wage premium over comparable roles without those skills, up from 25% the prior year. If you have demonstrable AI production experience — shipping models, building inference infrastructure, fine-tuning in a real system — that premium is directly applicable to your ask.

Real Numbers: The PwC 2025 Global AI Jobs Barometer analyzed close to 1 billion job postings across six continents and found that AI-skilled roles command a 56% wage premium. That's not a talking point. It's a live market data citation you can name in a negotiation call.

The anchor strategy is backed by negotiation research from Harvard's Program on Negotiation: setting a higher anchor number pulls the final settlement upward. The practical rule is to ask for 15% to 20% above your minimum acceptable number, not your ideal number. This gives you room to "concede" to something you would have happily accepted anyway, while the company feels they negotiated you down to a fair outcome. Both sides win.

Scripts for Every Moment in the Negotiation

This is the section most people need. Copy these, adjust the specifics, and use them verbatim if needed.

Asking for Time to Evaluate

You will almost always get this. Ask within hours of receiving the offer.

Script: "Thank you so much — I'm genuinely excited about this role and the team. Could I have until [specific date, typically 5 to 7 business days out] to review the full package with my family? I want to give this the consideration it deserves."

Never say "a few days." Name a specific date. It demonstrates you're organized and makes it harder for them to say no to something concrete.

The First Counter on Base Salary (No Competing Offer)

This is the core move. Lead with enthusiasm, cite data, name a specific number.

Script: "I'm very enthusiastic about joining [Company]. Based on my research into current market rates for senior ML engineers — particularly the premium for candidates with production AI experience — I was hoping we could discuss the base salary. The market data I'm seeing from Levels.fyi, combined with the 56% wage premium documented for AI-skilled roles in recent PwC research, suggests a base closer to [your anchor number] would reflect the current market. Is that a range we could explore?"

The First Counter When You Have a Competing Offer

A genuine competing offer is your strongest lever. Use it cleanly and without drama.

Script: "I want to be transparent with you because I'm genuinely more excited about [Company] than the other opportunity I'm considering. I have an offer from [Company B] at [total comp figure]. I'm not using that as a bluff — I'd prefer to be here — but I'd need [Company] to get closer to [specific target] to make the decision straightforward. Is there room to move on the total package?"

Never fabricate a competing offer. Recruiters in the same market talk. If the bluff is exposed, the offer is gone and your reputation in that network is damaged.

Common Mistake: Vaguely mentioning "other opportunities" without a specific number. This reads as a negotiating tactic, not a real alternative, and recruiters discount it immediately. A real competing offer has a company name and a dollar amount.

When They Say "This Is Our Best Offer"

This line ends negotiations for most candidates. It shouldn't.

Script: "I appreciate you being direct with me. I understand base is constrained by the band. Can we look at whether there's flexibility on the signing component or on the equity grant? I'd love to find a way to make this work."

Shifting the conversation from base to signing or equity after "best offer" is the single most effective tactic in this entire guide. The recruiter said base is their best — they did not say the total package is final.

Asking About Signing Bonus Specifically

Signing bonuses are a common tool companies use to close the distance between the band ceiling and what the candidate needs to accept. Recruiters often hold signing back to use as a negotiation move. Ask directly.

Script: "I understand base may be at the top of the band, which I respect. Is there flexibility on the signing component? A one-time signing bonus would help close the gap between the offer and what I need to make the transition financially comfortable, without changing the ongoing salary structure."

The Equity Ask

For roles at established companies with RSUs, equity is often the highest-value item in the package.

Script: "The equity in this offer is a meaningful part of the package. Could we discuss either increasing the initial RSU grant or structuring the first-year vest to accelerate slightly? Given the current AI talent market, I want to make sure we're structuring this in a way that reflects what I'm bringing to the role."

At senior levels, ask specifically about refresh grants. Annual refresh grants at FAANG companies typically run 25% to 75% of the initial grant depending on performance, but framing is set at hire. Ask: "What does the typical annual refresh cycle look like, and is there flexibility in the first-year grant to reflect market conditions?"

The Remote Work Addendum

Geography-based pay adjustments are real at many large companies. If you're being asked to take a pay cut for remote work or a lower-cost city, push back with data.

Script: "I noticed the offer reflects a [city] pay level. Given that I'll be working remotely and delivering the same output as a co-located team member, is there flexibility in applying a higher geo rate? The market data I'm seeing for this role doesn't differ as sharply by geography as the pay bands suggest."

The Competing Offer Strategy in Depth

A genuine competing offer is powerful but requires careful handling. The goal is to use it as a calibration tool, not a threat.

Name the company and the number. Vague "other interest" signals you're fishing. Specificity signals reality. Say: "I have an offer from Stripe at $280,000 total comp" — not "I'm exploring some other options."

Give them a real deadline tied to the other offer. "I need to respond to [Company B] by Friday" is a legitimate constraint that prompts action without manufactured pressure.

Where the strategy backfires: if your competing offer is from a company the interviewer respects more than their own, you may inadvertently show them you'd rather be elsewhere. Calibrate accordingly. A competing offer from OpenAI, Anthropic, or DeepMind carries different weight than one from a company nobody has heard of.

Pro Tip: If you don't have a competing offer but you're in late-stage interviews elsewhere, say that honestly: "I don't have a second offer yet, but I'm in final rounds with another company and expect to hear within two weeks." That's not a competing offer, but it's a real constraint, and it's true.

What Different Company Sizes Can and Cannot Move

Knowing where the flexibility lives saves you from pushing on doors that don't open.

LeverFAANG / Tier-1Mid-size TechStartup
Base salaryBand-constrained; needs VP approval to exceedModerate flexibilityHigh flexibility
Signing bonusHigh flexibility, $20K to $100K rangeModerate, $10K to $40K rangeLess common; equity trade-off instead
Equity (RSUs)High flexibility on quantityModerate; ISO/NSO structure variesVery high; options vs RSU negotiation
Annual bonus %Fixed formula, almost never negotiableSometimes adjustableOften tied to milestones
Remote/geo ratePolicy-driven, some flexibilityMore flexibleFully flexible
Start dateUsually flexible by 2 to 4 weeksFlexibleFlexible

Source: TeamRora AI/ML Salary Negotiation Guide, 2025; Levels.fyi community data

Handling the Exploding Offer

An exploding offer gives you 24 to 48 hours to accept. It's a pressure tactic. Companies use it to prevent you from completing other interviews and to avoid becoming a fallback option.

The right response is to ask for more time, directly and without apology.

Script: "Thank you for the offer — I'm very interested. I want to give this the serious consideration it deserves and make a decision I can fully commit to. Could we extend the deadline to [specific date, 5 to 7 days out]? I want to start with the right mindset, not a rushed one."

Most legitimate employers will grant this. If they refuse, consider what that tells you about how they treat employees on decisions that matter. An employer who won't give you a week to evaluate a multi-year career commitment is showing you their decision-making culture under pressure.

Key Insight: You can also use the exploding offer dynamic in reverse. If you're in final rounds with multiple companies, tell the one who moves fastest: "I have flexibility on timeline, but I do have competing processes that are likely to conclude within the next week. If you're able to expedite, I can make a decision quickly." This accelerates offers from your preferred company without lying.

When the Negotiation Is Over

Knowing when to stop is as important as knowing how to start. Here are the signals.

The negotiation is genuinely over when: the hiring manager (not just the recruiter) directly tells you the package is final, the company has made two moves in your direction and is showing fatigue, or you've received equity and signing offers that match or exceed your initial ask.

Walk away when: the company refuses to give you time to think, communication becomes cold or terse after a professional counter, the offer keeps changing structure in ways that make the total value unclear, or performance-based bonuses replace guaranteed compensation with no floor guarantee.

Accept when: the total compensation is within 5% of your target, the role's growth potential and team quality are compelling, and you can live with the number.

Red Flags in the Offer Itself

Before you negotiate, check that what you're negotiating over is real.

Watch for equity grants with no cliff: if there's no one-year cliff on vesting, the company can let you go at 11 months with zero equity. Standard vesting is 25% at the one-year cliff, then monthly after that.

Be cautious of performance bonuses with no documented formula. "Up to 20% bonus based on company and individual performance" with nothing in writing about how that's measured is a number that exists only on paper.

Watch for total compensation presentations that include projected equity value at current stock price. A $400,000 "total comp" package where $200,000 of that is unvested RSUs at today's price is half cash and half bet. Price that bet appropriately.

Common Mistake: Accepting equity grants denominated in shares rather than dollars without knowing the current valuation and 409A price (for startups). "50,000 options" means nothing without knowing what those options are worth and at what strike price. Always ask for the current share count, most recent 409A valuation, and the strike price.

Conclusion

Most AI and ML candidates leave money on the table not because they lack skill or credentials, but because the negotiation felt risky and they defaulted to acceptance. The data says otherwise: 73% of employers expect negotiation, the AI skills premium is documented at 56% above market (PwC, 2025), and signing bonuses at top tech companies routinely sit at $30,000 to $150,000 waiting for someone to ask.

The scripts in this article are calibrated to the AI/ML market specifically. They're professional, direct, and grounded in real market data rather than vague "believe in your worth" language that falls apart the moment a recruiter pushes back. Use the Levels.fyi data, name the PwC premium, and ask about each component of the package separately.

If you're building toward this kind of negotiation, it helps to be the candidate who warrants a strong counter. Our guide to AI and ML career fundamentals and our data scientist salary 2026 breakdown cover the skills and market conditions that make you credible in this conversation.

The worst outcome of a professional negotiation is that they say no. That almost never happens, and even when it does, you still have the job.

Career Q&A

What if I already accepted the offer verbally? Can I still negotiate?

A verbal acceptance without a signed offer letter is not binding in most US states. You can still negotiate, though you should acknowledge the situation directly: "I said I was excited to join, and I genuinely am — I just want to make sure we've fully explored the total package before I sign." Recruiters will usually re-engage. If the offer is already signed, your next negotiation window is your 6-month or 1-year review.

Should I negotiate every offer, even when I'm not sure of my position?

Yes, with one exception: if you're a first-time hire with no competing data points and the offer already matches the top of the Levels.fyi range for that role and level, negotiating risks a relationship for limited upside. In every other scenario, a polite, well-researched counter costs you nothing and frequently produces results. CareerBuilder data shows that 55% of candidates don't negotiate at all — which means the 45% who do have a measurable edge.

Can I negotiate remote work terms after the offer is made?

Absolutely. Remote work arrangements, home office stipends, equipment budgets, and even geography-based pay classifications are all negotiable at the offer stage. They become much harder to change once you're onboarded. If you want to work remotely from a lower-cost city but want the higher-market pay rate, the time to establish that is before you sign.

What do I say when asked "What's your current salary?" in a state where it's legal to ask?

Deflect first: "I'd like to understand the total compensation structure for this role before discussing a specific number." If they press, give a target range based on market data, not your current salary. Your current salary anchors their offer to your past employer's budget decisions, not to your value in this market. In California, New York, Colorado, Washington, Illinois, and several other states, asking for your salary history is now illegal — the employer must disclose the range first.

How do I negotiate when I'm changing industries or coming from a lower-paying role?

Frame the negotiation on the role's market rate, not on what you were previously paid. "Based on what senior ML engineers are earning for this scope of work, I'm targeting [number]" disconnects you from a previous salary that's irrelevant to the new context. Your previous comp is often the recruiter's anchor. Don't hand it to them.

Is it riskier to negotiate at a startup vs. a large company?

Different risk profile, not higher risk. At large companies, bands are formal and the recruiter knows exactly what's possible before you start. At startups, almost everything is negotiable — base, equity percentage, title, scope — but the equity math requires due diligence. Ask for the capitalization table and understand your dilution position before you negotiate on equity alone. A 0.5% stake in a company you can't value is not equivalent to 0.5% of a company with a known 409A.

What's the best way to negotiate a raise at a current employer using an outside offer?

Bring the offer in writing and be honest about your interest: "I received an offer from [Company] at [total comp]. I'm not actively looking to leave, but this made me realize I should understand where I stand here." Most managers will either match or give you the clearest possible signal that your comp trajectory isn't going to improve. Either outcome is information you need.

Sources

Practice with real Ad Tech data

90 SQL & Python problems · 15 industry datasets

250 free problems · No credit card

See all Ad Tech problems
Free Career Roadmaps8 PATHS

Step-by-step roadmaps from zero to job-ready — curated courses, salary data, and the exact learning order that gets you hired.

Explore all career paths