February 2018

Shail Jain, Knowledgent CEO,

     I’ve just gotten off the phone with Kerry Ann, your head of HR; given that she only said the investigation into my complaints is closed and that, “appropriate action has been taken,” I wanted to take the time to write a few notes. All of this was previously discussed with you in your office on 6 Feb 2018, with Tom Johnstone listening in on the phone for the better part of the conversation. Quotations are paraphrased unless noted as direct quote. I gave Gregory Banacki, Knowledgent General Counsel, advance notice of my intention to write this.

• During the meeting in your office, you readily agreed that Ari Yacobi, Knowledgent Chief Data Scientist, is not in fact a data scientist (AKA Adeel Arif, UPenn M.S. Tech Management 2014). You stated that you placed him in that role so that he could “put together a good team.” You did not respond when I asked why he has the title, “Chief Data Scientist.” You also suggested that he did not want me to present to you a Keras TensorFlow neural net image classifier I built before he could see it so that he did not “look dumb” (your words). As we discussed, he has been overseeing your “marquee” client project (your word) for six months with very poor results and a “data scientist” in charge who does not program. You were surprised to hear about the non-programming when I told you, talked to Ari immediately following our meeting, and then he came to me and asked me to take over the project because it was “getting nowhere.” I said no because they’ve already made a mess of it, and he deflated his shoulders and whined, “But (my name), what am I going to dooooo?” When he had to present at a conference in Boston (tweeted from Knowledgent account: “Our Chief Data Scientist Ari Yacobi presented our unique AI-based approach to molecular similarity search at the PRISME Forum’s semi-annual meeting in Cambridge, MA!”), he asked me the week before to do research for him and said that Knowledgent had never actually completed the project he was planning to present; when I asked which similarity algorithm he would tell conference attendees he used, he said he did not know. I asked him point blank how he was going to manage questions from people who actually know what they’re talking about, and he responded: “I don’t know, I’m a little nervous because I’ve never lied this much in a presentation before.” When I texted him the day after my meeting with you to say that I had just given you two weeks’ notice unless I don’t report to him anymore, he responded (direct quote, all [sic]): “I am little shocked and surprised - i took a lot of pride in having a person like you on my team and for finding you. Let’s talk tomorrow” I responded that I had nothing to say until I heard from you first. I have never seen any of his code, nor has he demonstrated he’s capable of giving any meaningful feedback on mine; based upon your admission that he’s “not a data scientist,” my question stands as to why he deserves the title. Tom Johnstone stopped responding to my e-mails when I expressed these concerns plainly to him. Why do you trust him after apparently misleading you for at least six months on your “marquee project”?
 Why is your senior leadership unwilling to communicate with employees who are actually capable? Tom, why do you talk to other people about what great work I was doing and then not respond to me?

• Sharad Kumar, Chief Technology Officer (CTO), drove a group of four additional colleagues (including Ari Yacobi, myself, another female coworker, and business analyst Prateek Peres-da-Silva) to an internal happy hour for a client project at a country club in NJ. On the way there, he said, “OK guys, we all know sexual harassment is wrong…. Me Too…. ok, don’t harass people…. but I mean really guys, if you’re not getting sexually harassed, there’s probably something wrong with you, right?” Ari said nothing. At the country club, Sharad came up to me at the bar and asked if there was “anyone special” in my life. I made up an answer. On the way back to Knowledgent HQ with the same people in the car, he repeated, “if you’re not getting sexually harassed, there’s probably something wrong with you.” I asked in a clear, loud voice if he says that to his daughter (whom I believe is 12 based on what he said in the interim). No one spoke. You brought up the fact that both you and Tom Johnstone have daughters; given that you currently oversee a company where this type of incident apparently does not faze senior management, my only response is sympathy to each of your daughters. 


• As you know, your sales men John Rubino and Manus Gallagher sent me to a client site before the Scope of Work (SOW) had been signed. Rubino asked me after my first day, in writing, to provide a list of duties I thought I might be performing there which could be added on the SOW. As you also know, Rubino and Manus also agreed to put my laptop specs in the SOW as I had previously had a very negative experience with 4G RAM on a different client machine. When the client admin contacted me and mentioned the laptop, I forwarded her message to the sales guys to confirm they had included specs in SOW. They said they left out specs because they don’t go in “this kind of SOW” and asked me to deal with client admin directly. Any company that makes data scientists talk to client admin about hardware without putting specs in contracts does not take data science seriously. I played you their apology voicemails which were left within 15 minutes of my Cc:ing Tom Johnstone in a response. Further, Rubino immediately began sending naïve-sounding e-mails to the client admin explaining machine spec requirements seemingly out of the blue. This is a multi-million dollar project. What do these guys get paid for? 


• Another one of your sales guys, Matt Arellano, didn’t know I was being allowed to listen in on the call at the time and said, “look, we all know the model is a piece of shit,” referring to a statistical model I was building. On subsequent calls, after repeated attempts to explain to him how to interpret classification reports, he continued to bungle accuracy and recall and complained that confusion matrices are “too confusing.” It is outside of his capacity to determine whether a model with 85.1% accuracy and 9% recall is a “piece of shit.” I was told by the project manager that Matt had asked me not to speak during the final client presentation. In a dry run with the client project lead, the project manager and client partner contradicted each other. The client turned to me; I had been silent. I calmly made eye contact with the two who had contradicted each other and then explained the results of my work. It was clear and made sense. He told our client partner afterward that he could tell something was not right because I hadn’t been reacting to anything anyone was saying. Fast forward: I met Matt Arellano for the first time in person on a trip to client site several months later to present the results to the client CIO. I answered many technical questions on the fly, including several about slides which he had requested be removed and against which I strongly cautioned. The Knowledgent client partner said he had never heard Matt Arellano so quiet during a meeting. I approached Matt Arellano afterward while he was saying good bye to the client, said good bye to him myself, and he turned and walked away without saying anything to me. This is in front of the client.


• The project manager from the paragraph above, in my first week at Knowledgent, said to me, “Even more people would listen to you if you wore high heels.” I responded, “I’m a programmer. I wear all black and flats. You wear high heels.”


• Several months later, Prateek Peres-da-Silva gave me blatant elevator eyes and asked if I had ever dressed for a client meeting before, in anticipation of the first one for my second project at Knowledgent. I knew he knew I had just been on client site for an 8-week POC, and said, “Yes, but go ahead.” He proceeded to point at my pants and tell me to wear dress pants (I was wearing dress pants); to point at my shirt and tell me to wear a button-up (I was wearing a 100% silk button-up); to point at my shoes and tell me to “wear shoes like (woman at the company who’s petite and nontrivially above me in age) wears…. basically, dress like she dresses.” I was wearing all black and black flats (leather, several hundred dollars), which I wear specifically because 1) we’re in NYC, Shail, and 2) I don’t want to deal with garbage like this trying to tell me what to wear when I’m a data scientist with serious work to do. After said meeting, he had car trouble and pulled over on the side of the highway. I asked him if he had a car jack; he said he thought so, took it out, and stared at it. I asked if he knew how to jack a car. I jacked his car for him. At some point he told me that Ari Yacobi had told him that I was the “only real data scientist at the company.”
 Additionally, as I mentioned to you, he said that in your office he announced to a group of people that the “marquee” client’s daughter is “hot.” He expressed concern that you would fire him afterward; however, it is my understanding that you took no action and continuously exhibit tacit approval of behavior like this.

• At a business lunch with Ari Yacobi and a “data scientist” (in quotes because, as you confirmed with Ari Yacobi after my meeting with you, this particular person does not program and would better serve in a business analyst role), those two were discussing their wives at home with their newborns. After a very long while of not participating in the conversation, I apologized for not having much to contribute as a means to move the conversation along. The “data scientist” eventually proceeded to ask me 1) how old i am, 2) if i want children, 3) if I’ve ever considered freezing my eggs. Ari Yacobi sat silently. I put my fork down and said, sternly, that I would like children and that is not something I would like to discuss at lunch. I have screenshots of glowing praise from this “data scientist” to me on Slack, both before and after this incident. 


• When I reported these things to your head of HR, Kerry Ann MacIsaac, after you left the phone call, she asked me, 1) why I did not report these things to her sooner; 2) in which month Prateek Peres-da-Silva scrutinized my outfit, since she had apparently done a dress code work shop a few months earlier (there is no dress code my outfit would have broken unless it were, “Do not wear all black”); 3) could not answer whether, on average, she thinks that men or women are sexually harassed more often in the United States, 4) why, if I thought Knowledgent did not take data science seriously, did I come and work there. I explained that a very competent female data scientist had interviewed me (whom, I have heard from Ari Yacobi, you lost under duress), and that Ari Yacobi willfully misrepresents his capacity as a data scientist. 
 In the phone conversation – in which I quit effective immediately because you insisted on keeping Ari Yacobi in his current position even though you did not argue when I said he’s a fraud – you even stooped to argue that, “‘data science’ means something different at Knowledgent.” You should probably share this alternative definition on the company website because right now Knowledgent makes pretty bold claims about how seriously it takes data science under the standard definition.

     Lastly, and this was not discussed in your office, I wanted to bring up your offer of having dinner at my choice of Italian restaurant in Manhattan (your terms) with you, Tom Johnstone, and Ari Yacobi if I could pull together a D3.js visualization within a set period of time. All four of us were present in the room for that offer. I’m curious why you thought I would somehow be interested in spending time with you and two married men outside of working hours instead of, for example, getting a bonus or a promotion or perhaps even a raise to meaningfully acknowledge that and all the other work I was doing for you. Would you have made the same offer to a hard-working male? Something tells me you would have thought a bonus were more appropriate. Maybe I’m wrong.

     Anyone at Ari Yacobi’s level or under would have throttled me in the upcoming performance review — what’s that? The Tableau guy who can’t program? The one with whom you said another data scientist also cancelled all monthly calls because they’re pointless and another demonstration of an ineffective and tortuous reporting system? The one who said if he needs “serious data science” done, he goes to the one you and I discussed can’t program? Please. — so I came straight to you with the concerns bulleted above. We know how that turned out.

     I think enough of my sentiment is implicit in what’s written above. I’ll end by saying that this industry, or any industry, does not have space for the kind of behavior you’ve been enabling.

     Best,

     REV

PS – Several people reached out to me since I quit, and it would be remiss of me not to say how much I appreciate them and miss them. You guys are great. Thank you.